Effects of Gravity and Non-Perpendicularity during Powder-Fed Directed Energy Deposition of Ni-Based Alloy 718 through Two Types of Coaxial Nozzle

The consequences of gravity and the nozzle inclination angle in the powder-fed Directed Energy Deposition (DED) process were examined in this study. We also sought to define guidelines and manufacturing strategies, depending on the DED system configuration and the nozzle type. To do so, two nozzle types were used: a continuous coaxial nozzle with a slit of 0.5 mm and a four-stream discrete coaxial nozzle. Although the main effects of the configurations and the nozzles are well-known, their effects on the clad characteristics and the deposition strategy are as yet unclear. In this paper, measurements of a single clad and the effects of different deposition strategies on cladding applications and inclined walls are presented, and the consequences for manufacturing processes are discussed. Based on a complete study of a single clad, working vertically, five different tilted deposition strategies were applied: three to a single clad and two to an inclined wall. The results for both the single clad and the inclined wall reflect a pattern of changes to height, width, area, and efficiency, at both small and large nozzle angles and deposition strategies. The inclined wall presents a maximum horizontal displacement that can be reached per layer, without geometrical distortions. The amount of material per layer has to be adapted to this limitation.


Introduction
Powder-fed Directed Energy Deposition (DED) technology is an Additive Manufacturing process in which a laser beam generates a melt pool on a metallic substrate onto which powder particles are injected. The added particles are fused and integrated into the melt pool before the whole melt pool solidifies, to form a high-quality metallurgical bond between the substrate and the added material. The design is formed layer by layer, thereby permitting the manufacture of complex geometrical parts.
This technology produces near-net-shape geometries that must be finished by a machining process if higher geometric accuracy and surface finish are required. Currently, the hybridization of additive manufacturing and machining processes combines these processes in a single multitask machine, increasing the productivity and the flexibility of the machining process.
Although the use of powder, as described by Toyserkani et al. [1], presents advantages (a wide range of available materials, the possibility of combining different alloys, high surface quality, and low susceptibility to cracking) over the use of wires, not all of the injected powder particles are deposited directly on the clad. Powder efficiency that entails both productive and economic consequences is

Machine
All tests were performed on an IBARMIA ZVH45/1600 Add+Process hybrid machine ( Figure 1). This multi-process machine (Ibarmia, Azkoitia, Spain) combines DED technology with a 5-axis milling and turning (horizontal and vertical) capability. It is equipped with a Precitec YC52 cladding head (Precitec, Gaggenau, Germany), a Sulzer Metco TWIN-10-C Powder Feeder (Oerlikon Metco, Freienbach, Switzerland), and a Yb-Fiber Rofin FL030 3 kW Laser generator (Coherent Rofin, Santa Clara, CA, USA) with a continuous wavelength of 1.07 µm. A 4-stream coaxial discrete nozzle and a coaxial continuous nozzle with a slit of 0.5 mm, both from Precitec, were used.

Experimental Testing
The experimental tests were divided into three different steps. First, single clads were produced under different processing conditions relating to laser Power (P), feed rate (v), and powder mass flow rate (ṁ p ), with the aim of ascertaining the optimum working conditions for Alloy 718 deposition. Once the best conditions had been established, the gravity-related effects of head tilting and non-perpendicular deposition were analyzed. The results obtained in this analysis were used to define two different strategies for manufacturing inclined walls.
All other processing variables were held constant during the experimentation, considering previous experience. The optimum working distance and laser spot diameter in both nozzles were obtained in the same way as Tabernero et al. [18], as described by Artaza et al. [19]. Their methodology consisted of using containers with different internal diameters for trapping the powder from the powder stream and for measuring the amount of powder collected in each container at different distances from the nozzle [2]. The optimum distance (the distance that collected a higher amount of powder) with that methodology was 11.5 mm for the continuous nozzle and 14.5 mm for the discrete one. Argon was employed as both a carrier and protective gas, at flow rates of 4.5 and 18 L·min −1 , respectively, using a laser spot size of 2.6 mm. The rotation percentage of the powder feeder defines different powder mass flow rates depending on the fluidity and the density of the powder material. The relation between the powder feeder rotating disk speed percentage and the powder mass flow rate was measured by weighing the powder collected in three minutes.

Experimental Testing
The experimental tests were divided into three different steps. First, single clads were produced under different processing conditions relating to laser Power (P), feed rate (v), and powder mass flow rate (ṁp), with the aim of ascertaining the optimum working conditions for Alloy 718 deposition. Once the best conditions had been established, the gravity-related effects of head tilting and nonperpendicular deposition were analyzed. The results obtained in this analysis were used to define two different strategies for manufacturing inclined walls.
All other processing variables were held constant during the experimentation, considering previous experience. The optimum working distance and laser spot diameter in both nozzles were obtained in the same way as Tabernero et al. [18], as described by Artaza et al. [19]. Their methodology consisted of using containers with different internal diameters for trapping the powder from the powder stream and for measuring the amount of powder collected in each container at different distances from the nozzle [2]. The optimum distance (the distance that collected a higher amount of powder) with that methodology was 11.5 mm for the continuous nozzle and 14.5 mm for the discrete one. Argon was employed as both a carrier and protective gas, at flow rates of 4.5 and 18 L·min −1 , respectively, using a laser spot size of 2.6 mm. The rotation percentage of the powder feeder defines different powder mass flow rates depending on the fluidity and the density of the powder material. The relation between the powder feeder rotating disk speed percentage and the powder mass flow rate was measured by weighing the powder collected in three minutes.

Single-Clad Testing
Clads of 60 mm in length were manufactured, to explore the optimum processing conditions of Alloy 718, by combining different laser power settings (P), feed rates (v), and powder mass flow rates (ṁ p ). The settings that were selected for each variable are shown in Table 2. These settings were selected on the basis of the conclusions obtained from previous works. As a result, a total of 72 clads were produced with each coaxial nozzle. In the analysis of the cladding, 4 cross-sections were selected at different positions, avoiding the first and the last 10 mm of the clads, thus avoiding process instabilities associated with start and stop conditions, as described by Ocelík et al. [20]. The cross-sections were analyzed in terms of porosity, height (h), width (w), area (A), and dilution depth (b), as shown in Figure 2. Samples were chemically etched by Kalling's N • 2 etching and measured with Motic SMZ-143 microscopy (Motic, Hong Kong, China) and Clemex Captiva®software (Clemex, Longueuil, QC, Canada). In addition, the Mass Deposition Rate (MDR), and the powder efficiency (ï p ) were calculated by using Equations (1) and (2), where ρ p is the density of the powder material. The mean value of the 4 sections was applied in models that link the characteristics of the single clad with the main process parameters (laser power, feed rate, and powder mass flow) that were studied. A similar method was followed by Kaiyun et al. [21], working with pre-placed AISI H13 powder, El Cheikh et al. [22] with 316 L powder, and De Oliveira et al. [23] with Nickel-Chromium based alloy powder. The shape of the clad, the process variability of each characteristic, and the error of the empirical models were also analyzed.

Process Parameter
Level P (W) 1900, 2200, and 2500 v (mm·min −1 ) 300, 500, 700, 900, 1100, and 1300 ṁp (g·min −1 ) 11, 14, 18, and 21 Type of coaxial nozzle Continuous (0.5 mm slit), discrete (4-stream) In the analysis of the cladding, 4 cross-sections were selected at different positions, avoiding the first and the last 10 mm of the clads, thus avoiding process instabilities associated with start and stop conditions, as described by Ocelík et al. [20]. The cross-sections were analyzed in terms of porosity, height (h), width (w), area (A), and dilution depth (b), as shown in Figure 2. Samples were chemically etched by Kalling's N°2 etching and measured with Motic SMZ-143 microscopy (Motic, Hong Kong, China) and Clemex Captiva® software (Clemex, Longueuil, QC, Canada). In addition, the Mass Deposition Rate (MDR), and the powder efficiency (ɳp) were calculated by using Equations (1) and (2), where ρp is the density of the powder material. The mean value of the 4 sections was applied in models that link the characteristics of the single clad with the main process parameters (laser power, feed rate, and powder mass flow) that were studied. A similar method was followed by Kaiyun et al. [21], working with pre-placed AISI H13 powder, El Cheikh et al. [22] with 316 L powder, and De Oliveira et al. [23] with Nickel-Chromium based alloy powder. The shape of the clad, the process variability of each characteristic, and the error of the empirical models were also analyzed.

Analysis of the Effects of Head Tilting and Non-Perpendicular Deposition
Testing the single clad helped to define the two optimum deposition conditions for Alloy 718 when employing both the continuous and the discrete coaxial nozzles. The following conditions were considered for the selection: a defect-free cross-section, a Mass Deposition Rate (MDR) higher than 0.5 Kg·h −1 , maximum powder efficiency, and an Aspect Ratio (the relationship between clad width and height) between 3 and 5. The latter is related to the appearance of porosity and cracking in overlapping tracks, according to Cárcel et al. [24].
The selected processing conditions were used to perform single tests, using three different strategies to analyze the gravitational effects of head tilting and the effects of non-perpendicular deposition. The angles α and β were defined for that purpose: α represents the tilting angle of the cladding head in relation to the vertical axis, and β represents the inclination angle of the substrate on the horizontal axis. The three strategies can be described as follows:

Analysis of the Effects of Head Tilting and Non-Perpendicular Deposition
Testing the single clad helped to define the two optimum deposition conditions for Alloy 718 when employing both the continuous and the discrete coaxial nozzles. The following conditions were considered for the selection: a defect-free cross-section, a Mass Deposition Rate (MDR) higher than 0.5 kg·h −1 , maximum powder efficiency, and an Aspect Ratio (the relationship between clad width and height) between 3 and 5. The latter is related to the appearance of porosity and cracking in overlapping tracks, according to Cárcel et al. [24].
The selected processing conditions were used to perform single tests, using three different strategies to analyze the gravitational effects of head tilting and the effects of non-perpendicular deposition. The angles α and β were defined for that purpose: α represents the tilting angle of the cladding head in relation to the vertical axis, and β represents the inclination angle of the substrate on the horizontal axis. The three strategies can be described as follows: 1.
Strategy 1 (α = β, Figure 3a): the cladding head is tilted and deposition is perpendicular. Thus, in this case, the substrate is inclined, so that perpendicularity is maintained between the head and the substrate, and the head-tilting factor (i.e., gravity) influences the clad characteristics.

2.
Strategy 2 (α = 0, Figure 3b): the cladding head is held vertical and deposition is performed on an inclined substrate. In this strategy, non-perpendicularity is the factor that influences the clad characteristics, since gravity has no effect on the powder stream.

3.
Strategy 3 (β = 0, Figure 3c): a combination of both effects (tilting and non-perpendicularity). As the combination of Strategy 1 and Strategy 2, it consists of tilting the cladding head on a horizontal substrate. Both the tilting head and non-perpendicularity influence the clad characteristics.
Metals 2020, 10, x FOR PEER REVIEW 6 of 21 1. Strategy 1 (α = β, Figure 3a): the cladding head is tilted and deposition is perpendicular. Thus, in this case, the substrate is inclined, so that perpendicularity is maintained between the head and the substrate, and the head-tilting factor (i.e., gravity) influences the clad characteristics. 2. Strategy 2 (α = 0, Figure 3b): the cladding head is held vertical and deposition is performed on an inclined substrate. In this strategy, non-perpendicularity is the factor that influences the clad characteristics, since gravity has no effect on the powder stream. 3. Strategy 3 (β = 0, Figure 3c): a combination of both effects (tilting and non-perpendicularity). As the combination of Strategy 1 and Strategy 2, it consists of tilting the cladding head on a horizontal substrate. Both the tilting head and non-perpendicularity influence the clad characteristics.
(a) (b) (c) The processing parameters for the study are presented listed in Table 3. As in the previous section, 4 cross-sections of the clads were analyzed, using the same methodology. Finally, the symmetry of the clads was analyzed, using different strategies and different α and β angles. The cross-sections were compared to an empirical model of the cross-sectional area, obtained in Section 2.3.1, as a reference clad to measure symmetry. The model compares the cross-section with a seconddegree parabola or a mix of a second-degree parabola and an ellipse, depending on the value of the clad area. The difference between the clad area on both sides of the axis of symmetry of the reference clad was calculated. Table 3. Process parameters for the analysis of the effect of tilting and non-perpendicular deposition.

Manufacturing of Inclined Walls
The objective of this section is to extend the results obtained in the previous sections for a single clad to the generic case of manufacturing inclined walls on a horizontal substrate (β = 0), which requires multilayer deposits of over one clad per layer. To do so, two strategies were defined ( Figure  4): The processing parameters for the study are presented listed in Table 3. As in the previous section, 4 cross-sections of the clads were analyzed, using the same methodology. Finally, the symmetry of the clads was analyzed, using different strategies and different α and β angles. The cross-sections were compared to an empirical model of the cross-sectional area, obtained in Section 2.3.1, as a reference clad to measure symmetry. The model compares the cross-section with a second-degree parabola or a mix of a second-degree parabola and an ellipse, depending on the value of the clad area. The difference between the clad area on both sides of the axis of symmetry of the reference clad was calculated.

Manufacturing of Inclined Walls
The objective of this section is to extend the results obtained in the previous sections for a single clad to the generic case of manufacturing inclined walls on a horizontal substrate (β = 0), which requires multilayer deposits of over one clad per layer. To do so, two strategies were defined ( Figure 4):

1.
Strategy A (4-axis, Figure 4a): the cladding head is tilted in accordance with the inclination of the wall (θ = 90 − α). Using this strategy, the material is applied in the direction of the wall.

2.
Strategy B (3-axis, Figure 4b): the cladding head is maintained in a vertical position (α = 0). Using this strategy, the material is vertically applied.
The option of maintaining a perpendicular head and tilting the substrate to manufacture an inclined wall is the easiest option for the multilayer manufacturing of walls with no loss of shape, 1. Strategy A (4-axis, Figure 4a): the cladding head is tilted in accordance with the inclination of the wall (θ = 90 − α). Using this strategy, the material is applied in the direction of the wall. 2. Strategy B (3-axis, Figure 4b): the cladding head is maintained in a vertical position (α = 0). Using this strategy, the material is vertically applied. The option of maintaining a perpendicular head and tilting the substrate to manufacture an inclined wall is the easiest option for the multilayer manufacturing of walls with no loss of shape, as was demonstrated by Möller et al. [16]. However, that option requires a tilting table, which was not a feature of the hybrid machine used for this study.  The angle of the wall was obtained by defining the horizontal and vertical displacement (ΔX and ΔZ, respectively) in the tool path for each layer. These two displacements were calculated (Equations (3) and (7)) with the desired wall angle (θ), which is related to angle α (from the study of Section 2.3.2) in Strategy A, and the layer height (HL). The latter was calculated (Equation (4)) by dividing the total area of the layer (AT) by the sum of the total width (WT) and horizontal displacement (ΔX), supposing that the material of each layer covers all the area needed to obtain a perfect wall with no waviness (Figure 5a). The total area and the total width of the layer were calculated by using Equations (5) and (6), respectively, where NC is the number of single clads per layer. The area (A) and the width (w) of the single clad were calculated by using the results obtained in Section 2.3.2 for each specific α and β condition, where do is the percentage overlap of the adjacent clads, and K is a correction factor that takes into account the increase of the total width with the number of layers, until a stable width is reached. The phenomenon, shown in Figure 5b, occurs due to the displacement of the melt pool caused by the surface shape of the previous layer and the increased temperature of the wall, and can be reduced by controlling the laser power depending on the temperature measurement, as described by Möller et al. [16]. The number of layers of each wall (NL) was 30.
The process parameters for the laser-beam multilayer deposits to build the walls are shown in Table 4. The overlap percentage and the dwell time were selected from previous works. To evaluate the effects of inclination on the geometry of the wall, the distance between the center lines of the two overlapped clads was in all cases fixed at the same value as for a vertically manufactured wall.

∆ =
(3) The angle of the wall was obtained by defining the horizontal and vertical displacement (∆X and ∆Z, respectively) in the tool path for each layer. These two displacements were calculated (Equations (3) and (7)) with the desired wall angle (θ), which is related to angle α (from the study of Section 2.3.2) in Strategy A, and the layer height (H L ). The latter was calculated (Equation (4)) by dividing the total area of the layer (A T ) by the sum of the total width (W T ) and horizontal displacement (∆X), supposing that the material of each layer covers all the area needed to obtain a perfect wall with no waviness (Figure 5a). The total area and the total width of the layer were calculated by using Equations (5) and (6), respectively, where N C is the number of single clads per layer. The area (A) and the width (w) of the single clad were calculated by using the results obtained in Section 2.3.2 for each specific α and β condition, where d o is the percentage overlap of the adjacent clads, and K is a correction factor that takes into account the increase of the total width with the number of layers, until a stable width is reached. The phenomenon, shown in Figure 5b, occurs due to the displacement of the melt pool caused by the surface shape of the previous layer and the increased temperature of the wall, and can be reduced by controlling the laser power depending on the temperature measurement, as described by Möller et al. [16]. The number of layers of each wall (N L ) was 30.
The process parameters for the laser-beam multilayer deposits to build the walls are shown in Table 4. The overlap percentage and the dwell time were selected from previous works. To evaluate the effects of inclination on the geometry of the wall, the distance between the center lines of the two overlapped clads was in all cases fixed at the same value as for a vertically manufactured wall.
Metals 2020, 10, 560  Although the intention was to test both strategies with both types of coaxial nozzles (continuous and discrete), the continuous nozzle was only used in Strategy B, due to the poor results obtained in the analysis of the effects of head tilting and non-perpendicular deposition (Section 2.3.2).
The results were analyzed in terms of geometry, powder efficiency, and MDR. In this case, the powder efficiency was calculated by weighing the substrate before and after deposition (mi and mf, respectively) and using Equation (8), where tDED is the processing time of the DED process. The MDR was calculated by using Equation (2). Finally, the geometry of the wall was evaluated by the total width (WW), the wall height (HW), the initial width of the wall (Wi), which is the width of the wall before the increase of the width K, and the length (LW) of the wall ( Figure 6). These dimensions were measured on a cross-section of the wall, with the Clemex Captiva® software. The measured values were compared to the theoretical values obtained with Equations (9)-(12).  Although the intention was to test both strategies with both types of coaxial nozzles (continuous and discrete), the continuous nozzle was only used in Strategy B, due to the poor results obtained in the analysis of the effects of head tilting and non-perpendicular deposition (Section 2.3.2).
The results were analyzed in terms of geometry, powder efficiency, and MDR. In this case, the powder efficiency was calculated by weighing the substrate before and after deposition (m i and m f , respectively) and using Equation (8), where t DED is the processing time of the DED process. The MDR was calculated by using Equation (2). Finally, the geometry of the wall was evaluated by the total width (W W ), the wall height (H W ), the initial width of the wall (W i ), which is the width of the wall before the increase of the width K, and the length (L W ) of the wall ( Figure 6). These dimensions were measured on a cross-section of the wall, with the Clemex Captiva®software. The measured values were compared to the theoretical values obtained with Equations (9)- (12).
Metals 2020, 10, 560 9 of 20 = sin (12) Figure 6. Geometric dimensions of the wall. Table 5 shows the parameters of the clads that met the requirements presented in Section 2.1 and that were considered suitable for analyzing the effects of gravity and non-perpendicularity. The average values of the four cross-sections of all the clads were used to obtain equations of the single-clad characteristics (width, height, area, and dilution depth), according to the main parameters (laser power, feed rate, and powder mass flow) in the range under study. These equations were obtained by a second-degree polynomial regression and can be used to calculate the parameters needed for manufacturing a single clad with a specific characteristic. Figure 7 shows the evolution of the height and the width at a fixed powder mass flow rate of 14 g·min −1 for both nozzles. Similar trends were observed in both nozzles for these two characteristics. Laser power had no significant  Table 5 shows the parameters of the clads that met the requirements presented in Section 2.1 and that were considered suitable for analyzing the effects of gravity and non-perpendicularity. The average values of the four cross-sections of all the clads were used to obtain equations of the single-clad characteristics (width, height, area, and dilution depth), according to the main parameters (laser power, feed rate, and powder mass flow) in the range under study. These equations were obtained by a second-degree polynomial regression and can be used to calculate the parameters needed for manufacturing a single clad with a specific characteristic. Figure 7 shows the evolution of the height and the width at a fixed powder mass flow rate of 14 g·min −1 for both nozzles. Similar trends were observed in both nozzles for these two characteristics. Laser power had no significant effect on the height, although the width increased at higher powers. As expected, both height and width decreased as the feed rate increased.

Single-Clad Testing
Metals 2020, 10, x FOR PEER REVIEW 10 of 21 effect on the height, although the width increased at higher powers. As expected, both height and width decreased as the feed rate increased. The maximum values for powder efficiency and the MDR were achieved at a laser power of 2500 W for both nozzles ( Figure 8). This is due to the increase of the melt pool size at higher laser power. It must be pointed out that if the energy input becomes excessive the efficiency could be reduced by ablation of the injected powder particles as modeled by Volpp et al. [25]. This possibility has to be considered in recent processes like Extreme High-Speed Laser Material Deposition (EHLA) [26]. In the case of powder efficiency, different behaviors were observed. While the four-stream nozzle showed an optimum powder mass flow rate that optimized powder efficiency and varied with the feed rate, the effect of the powder mass flow rate on powder efficiency was not significant in the case of the continuous nozzle. On the other hand, when increasing the feed rate, both powder efficiency and MDR decreased in both nozzles. This is due to the decrease of the size of the melt pool during deposition on the substrate, making the powder catchment lower, as concluded Prasad et al., using a green laser on copper [27]. Finally, the maximum powder efficiency and the maximum MDR for the four-stream nozzle and for the continuous nozzle amounted to 64% and 0.75 kg·h −1 and 90% and 1.04 kg·h −1 , respectively. The maximum values for powder efficiency and the MDR were achieved at a laser power of 2500 W for both nozzles (Figure 8). This is due to the increase of the melt pool size at higher laser power. It must be pointed out that if the energy input becomes excessive the efficiency could be reduced by ablation of the injected powder particles as modeled by Volpp et al. [25]. This possibility has to be considered in recent processes like Extreme High-Speed Laser Material Deposition (EHLA) [26]. In the case of powder efficiency, different behaviors were observed. While the four-stream nozzle showed an optimum powder mass flow rate that optimized powder efficiency and varied with the feed rate, the effect of the powder mass flow rate on powder efficiency was not significant in the case of the continuous nozzle. On the other hand, when increasing the feed rate, both powder efficiency and MDR decreased in both nozzles. This is due to the decrease of the size of the melt pool during deposition on the substrate, making the powder catchment lower, as concluded Prasad et al., using a green laser on copper [27]. Finally, the maximum powder efficiency and the maximum MDR for the four-stream nozzle and for the continuous nozzle amounted to 64% and 0.75 kg·h −1 and 90% and 1.04 kg·h −1 , respectively. Metals 2020, 10, x FOR PEER REVIEW 11 of 21 Finally, with regard to the shape of the clad, the average value of the clad area was between the value of a second-degree parabola and an ellipse ( Figure 9) and was closer to the second-degree parabola when the area was smaller than 3 mm². With regard to penetration, the average value of the penetration area was also between the value of a second-degree parabola and an ellipse and was closer to a second-degree parabola when smaller than 3.5 mm².  Finally, with regard to the shape of the clad, the average value of the clad area was between the value of a second-degree parabola and an ellipse ( Figure 9) and was closer to the second-degree parabola when the area was smaller than 3 mm 2 . With regard to dilution, the average value of the dilution area was also between the value of a second-degree parabola and an ellipse and was closer to a second-degree parabola when smaller than 3.5 mm 2 . Finally, with regard to the shape of the clad, the average value of the clad area was between the value of a second-degree parabola and an ellipse ( Figure 9) and was closer to the second-degree parabola when the area was smaller than 3 mm². With regard to penetration, the average value of the penetration area was also between the value of a second-degree parabola and an ellipse and was closer to a second-degree parabola when smaller than 3.5 mm².

Analysis of the Effects of Head Tilting and Non-Perpendicular Deposition
The results of the geometrical dimensions of the clad (h, w, and A) obtained using the different strategies and the different coaxial nozzles at a feed rate of 500 mm·min −1 are shown in Figure 10. Those dimensions are presented in comparison with the control clad characterized in Section 3.1 (α = 0, β = 0) at different angles. The graphs of each area are proportional to the MDR and the efficiency rate (see Equations (1) and (2)). Similar results were obtained at a feed rate of 900 mm·min −1 , as can be seen in Figure 11.

Analysis of the Effects of Head Tilting and Non-Perpendicular Deposition
The results of the geometrical dimensions of the clad (h, w, and A) obtained using the different strategies and the different coaxial nozzles at a feed rate of 500 mm·min −1 are shown in Figure 10. Those dimensions are presented in comparison with the control clad characterized in Section 3.1 (α = 0, β = 0) at different angles. The graphs of each area are proportional to the MDR and the efficiency rate (see Equations (1) and (2)). Similar results were obtained at a feed rate of 900 mm·min −1 , as can be seen in Figure 11.

Analysis of the Effects of Head Tilting and Non-Perpendicular Deposition
The results of the geometrical dimensions of the clad (h, w, and A) obtained using the different strategies and the different coaxial nozzles at a feed rate of 500 mm·min −1 are shown in Figure 10. Those dimensions are presented in comparison with the control clad characterized in Section 3.1 (α = 0, β = 0) at different angles. The graphs of each area are proportional to the MDR and the efficiency rate (see Equations (1) and (2)). Similar results were obtained at a feed rate of 900 mm·min −1 , as can be seen in Figure 11.  As expected, no significant gravitational effects were observed when tilting the four-stream coaxial nozzle (Strategy 1, α = β). On the contrary, when employing the continuous coaxial nozzle, the clad height was lowered to 87.3% of the height of the control clad at a tilt angle of 5 • and to 30.1% at a tilt angle of 30 • . Nevertheless, the clad width remained constant at the same feed rate and powder mass flow, as it depends mainly on the spot size and the laser power, which were both constant in this case. As a result, the area was directly proportional to the width and the height (as observed in the Section 3.1), and both the MDR and powder efficiency reflected a very similar proportional relationship with the control clad. The area, the MDR, and powder efficiency were therefore all reduced to 88.5% of the control clad, at a tilt angle of 5 • , and to 27.3%, at a tilt angle of 30 • (Figure 11). This dimensional reduction, due to the deviation of the powder stream, was caused by gravitational effects. Whereas in the four-stream coaxial nozzle, the powder was injected by four individual pipes with a diameter of 2 mm, forming a powder stream focus, in the continuous coaxial nozzle, the powder was fed into a ring-shaped expansion chamber inside the nozzle, reducing the flow velocity. In this chamber, a homogeneous "powder cloud" was formed which was then fed into a coned-shaped slit of 0.5 mm, leaving the nozzle in the form of a hollow cone. When tilting the continuous coaxial nozzle, the gravity affected the powder homogeneous distribution obtained in the expansion chamber during was fed into the coned-shaped slit, deflecting the powder to the same side (Figure 12). At a higher tilting angle, the deflected amount of powder increased, as did the disturbances in the powder stream caused by the interaction of the powder particles with each other on the same side of the slit. Therefore, it changed the powder stream with the continuous coaxial nozzle functioning more as a lateral nozzle [6] with high divergence, causing a scattered distribution of the powder in the melt pool.
Metals 2020, 10, x FOR PEER REVIEW 13 of 21 Figure 11. Evolution of the clad area in relation to α and β at 500 and 900 mm·min −1 .
As expected, no significant gravitational effects were observed when tilting the four-stream coaxial nozzle (Strategy 1, α = β). On the contrary, when employing the continuous coaxial nozzle, the clad height was lowered to 87.3% of the height of the control clad at a tilt angle of 5° and to 30.1% at a tilt angle of 30°. Nevertheless, the clad width remained constant at the same feed rate and powder mass flow, as it depends mainly on the spot size and the laser power, which were both constant in this case. As a result, the area was directly proportional to the width and the height (as observed in the Section 3.1), and both the MDR and powder efficiency reflected a very similar proportional relationship with the control clad. The area, the MDR, and powder efficiency were therefore all reduced to 88.5% of the control clad, at a tilt angle of 5°, and to 27.3%, at a tilt angle of 30° ( Figure 11). This dimensional reduction, due to the deviation of the powder stream, was caused by gravitational effects. Whereas in the four-stream coaxial nozzle, the powder was injected by four individual pipes with a diameter of 2 mm, forming a powder stream focus, in the continuous coaxial nozzle, the powder was fed into a ring-shaped expansion chamber inside the nozzle, reducing the flow velocity. In this chamber, a homogeneous "powder cloud" was formed which was then fed into a conedshaped slit of 0.5 mm, leaving the nozzle in the form of a hollow cone. When tilting the continuous coaxial nozzle, the gravity affected the powder homogeneous distribution obtained in the expansion chamber during was fed into the coned-shaped slit, deflecting the powder to the same side ( Figure  12). At a higher tilting angle, the deflected amount of powder increased, as did the disturbances in the powder stream caused by the interaction of the powder particles with each other on the same side of the slit. Therefore, it changed the powder stream with the continuous coaxial nozzle functioning more as a lateral nozzle [6] with high divergence, causing a scattered distribution of the powder in the melt pool. A different behavior was observed when employing Strategy 2 (α = 0). In this case, clad height presented a similar trend in both nozzles: a lower height at a larger inclined substrate angle β. It was reduced to 76.7% and to 70% of the control clad, at 30°, with the four-stream nozzle and the continuous nozzle, respectively. In contrast, the clad width as a percentage of the control clad increased as the non-perpendicularity increased, reaching 114.3% and 125.1%, at 30°, for the fourstream nozzle and the continuous nozzle, respectively. This increase occurs due to the elliptical projection of the laser spot on the inclined substrate, which widens the melt pool, and therefore widens the surface on which the powder can be trapped. However, only the continuous coaxial nozzle showed a slight area increases at angles lower than 15° and, consequently, an increase of the efficiency and the MDR. The area, the MDR, and powder efficiency remained constant at angles lower than 20° with the four-stream nozzle. Therefore, a wider melt pool showed a weak effect on the powder catchment that needs to be investigated in further studies. At angles higher than 20°, the area was reduced to 82.8% and 85.8%, at 30°, for the four-stream nozzle and the continuous nozzle, respectively. This reduction occurs because most of the inclined melt pool, where the material was deposited, was off the optimum working distance (where the powder mass flow converges [2,13]). A different behavior was observed when employing Strategy 2 (α = 0). In this case, clad height presented a similar trend in both nozzles: a lower height at a larger inclined substrate angle β. It was reduced to 76.7% and to 70% of the control clad, at 30 • , with the four-stream nozzle and the continuous nozzle, respectively. In contrast, the clad width as a percentage of the control clad increased as the non-perpendicularity increased, reaching 114.3% and 125.1%, at 30 • , for the four-stream nozzle and the continuous nozzle, respectively. This increase occurs due to the elliptical projection of the laser spot on the inclined substrate, which widens the melt pool, and therefore widens the surface on which the powder can be trapped. However, only the continuous coaxial nozzle showed a slight area increases at angles lower than 15 • and, consequently, an increase of the efficiency and the MDR. The area, the MDR, and powder efficiency remained constant at angles lower than 20 • with the four-stream nozzle. Therefore, a wider melt pool showed a weak effect on the powder catchment that needs to be investigated in further studies. At angles higher than 20 • , the area was reduced to 82.8% and 85.8%, at 30 • , for the four-stream nozzle and the continuous nozzle, respectively. This reduction occurs because most of the inclined melt pool, where the material was deposited, was off the optimum working distance (where the powder mass flow converges [2,13]). Up until 20 • , the reduction in height was related to the surface tension of the melt pool that depended on the balance between several forces, such as the Marangoni force, as explained by Egry et al. [28]. The molten material flow inside the melt pool not only influences its final geometry, but it also affects the reached maximum temperatures, as simulated Arrizubieta et al. in their model of the melt-pool dynamics [29]. In this case, the width widened while the area remained unchanged. Thus, the shape of the melt pool changed to one with less surface area, due to the reduction in height. At angles larger than 20 • , the reduction in height was related to the combined effect of surface tension and reduced powder deposition. This change of shape of the melt pool increased the aspect ratio and decreased the wetting angle of the clad (Figure 13). The latter related to the increase of adhesive force to the substrate.
The results obtained when employing Strategy 3 (β = 0) can be explained by the results obtained with the other two strategies, as it combines the gravitational effect of head tilting (Strategy 1) with the effect of non-perpendicularity (Strategy 2). In the case of Strategy 3, both clad height and clad width were affected by non-perpendicularity when using the four-stream nozzle, and by a combination of both effects when using the continuous nozzle. In the case of the four-stream nozzle, the effects of Strategy 2 and Strategy 3 were similar, although with some differences. Strategy 2 reduced the clad area to 82.8% of the control clad, and Strategy 3 reduced the area to 91.1%. While those results are similar, they point to some differences in the amount of powder deposits that merit further study. In the case of the continuous nozzle, Strategy 3 combined the reduced height of Strategy 1 with the wider width of Strategy 2. As a result, the area that Strategy 3 generated was larger than the area generated by Strategy 1, up until an angle of 20 • . At higher angles, the increased width due to non-perpendicularity would be insufficient to compensate for the deviation of the powder mass flow, due to nozzle tilting. Thus, the height and the area of the clads were lower than in Strategy 1. As a result, the area, the MDR, and the powder efficiency varied only slightly when the four-stream nozzle head was tilted, while those same parameters were reduced to 17.5% of the control clad with the continuous nozzle.
Finally, the symmetry of the clad area was observed not to differ from the single-clad study, except for the clads that followed Strategy 1 at 30 • (Figure 13). In that case, with a feed rate of 500 mm·min −1 , the percentage of the total area separated by the symmetry axis was 42.7% and 57.3%, respectively, where the higher area was deposited in the direction of gravity. In that case, the peak point shifting studied by Hao et al. [17] was obvious. Up until 20°, the reduction in height was related to the surface tension of the melt pool that depended on the balance between several forces, such as the Marangoni force, as explained by Egry et al. [28]. The molten material flow inside the melt pool not only influences its final geometry, but it also affects the reached maximum temperatures, as simulated Arrizubieta et al. in their model of the melt-pool dynamics [29]. In this case, the width widened while the area remained unchanged. Thus, the shape of the melt pool changed to one with less surface area, due to the reduction in height. At angles larger than 20°, the reduction in height was related to the combined effect of surface tension and reduced powder deposition. This change of shape of the melt pool increased the aspect ratio and decreased the wetting angle of the clad (Figure 13). The latter related to the increase of adhesive force to the substrate.
The results obtained when employing Strategy 3 (β = 0) can be explained by the results obtained with the other two strategies, as it combines the gravitational effect of head tilting (Strategy 1) with the effect of non-perpendicularity (Strategy 2). In the case of Strategy 3, both clad height and clad width were affected by non-perpendicularity when using the four-stream nozzle, and by a combination of both effects when using the continuous nozzle. In the case of the four-stream nozzle, the effects of Strategy 2 and Strategy 3 were similar, although with some differences. Strategy 2 reduced the clad area to 82.8% of the control clad, and Strategy 3 reduced the area to 91.1%. While those results are similar, they point to some differences in the amount of powder deposits that merit further study. In the case of the continuous nozzle, Strategy 3 combined the reduced height of Strategy 1 with the wider width of Strategy 2. As a result, the area that Strategy 3 generated was larger than the area generated by Strategy 1, up until an angle of 20°. At higher angles, the increased width due to non-perpendicularity would be insufficient to compensate for the deviation of the powder mass flow, due to nozzle tilting. Thus, the height and the area of the clads were lower than in Strategy 1. As a result, the area, the MDR, and the powder efficiency varied only slightly when the four-stream nozzle head was tilted, while those same parameters were reduced to 17.5% of the control clad with the continuous nozzle.
Finally, the symmetry of the clad area was observed not to differ from the single-clad study, except for the clads that followed Strategy 1 at 30° (Figure 13). In that case, with a feed rate of 500 mm·min −1 , the percentage of the total area separated by the symmetry axis was 42.7% and 57.3%, respectively, where the higher area was deposited in the direction of gravity. In that case, the peak point shifting studied by Hao et al. [17] was obvious.

Control 0°
Strategy 1 at 30° Strategy 2 at 30° Strategy 3 at 30° Figure 13. Comparison of the clad shapes (blue color) with the empirical model (red line) when using a four-stream nozzle at different strategies and angles, at v = 500 mm·min −1 . The aspect ratio increases, and the wetting angle decreased with Strategy 2 and Strategy 3.

Manufacturing of Inclined Walls
Although walls could be generated at angles between 60° and 90° with the four-stream nozzle and Strategy A (Figure 14), the walls were unsuccessful at angles smaller than 80° when using Strategy B at 500 mm·min −1 . Table 6 summarizes the results for powder efficiency and the MDR when manufacturing inclined walls using the different strategies and the different nozzle types. Figure 13. Comparison of the clad shapes (blue color) with the empirical model (red line) when using a four-stream nozzle at different strategies and angles, at v = 500 mm·min −1 . The aspect ratio increases, and the wetting angle decreased with Strategy 2 and Strategy 3.

Manufacturing of Inclined Walls
Although walls could be generated at angles between 60 • and 90 • with the four-stream nozzle and Strategy A (Figure 14), the walls were unsuccessful at angles smaller than 80 • when using Strategy B at 500 mm·min −1 . Table 6 summarizes the results for powder efficiency and the MDR when manufacturing inclined walls using the different strategies and the different nozzle types.
Compared to the single-clad case, at 90 • , the efficiency decreased with both nozzles (4-6% less). The maximum powder efficiency and MDR values were obtained with the continuous nozzle employing Strategy B, but it was not possible to manufacture walls without defects at angles lower than 80 • . Strategy A, with the four-stream nozzle, in all cases generated walls with similar powder efficiency and MDRs to the single-clad characteristics resulting from the analysis of the effects of head tilting and non-perpendicular deposition. Compared to the single-clad case, at 90°, the efficiency decreased with both nozzles (4-6% less). The maximum powder efficiency and MDR values were obtained with the continuous nozzle employing Strategy B, but it was not possible to manufacture walls without defects at angles lower than 80°. Strategy A, with the four-stream nozzle, in all cases generated walls with similar powder efficiency and MDRs to the single-clad characteristics resulting from the analysis of the effects of head tilting and non-perpendicular deposition.
The deterioration of the walls with Strategy B at angles smaller than 80° was due to less horizontal clad deposition per layer than the expected ΔX. The material deposited on the edge of the wall has to maintain the shape with low distortion for correct multilayering. The shape of the molten material depends on whether the surface tension and viscosity of the molten material at the DED process temperature are sufficient to withstand the shear stress produced by the gravitational forces. In addition, the time during which gravity can shape the molten material depends on the cooling rate that is, in turn, defined by each strategy.  Table 6. Efficiency and MDR results obtained with different nozzles, strategies, and wall angles (note that the walls that failed are marked with an "X" and walls obtained at 900 mm·min −1 are marked with *). The deterioration of the walls with Strategy B at angles smaller than 80 • was due to less horizontal clad deposition per layer than the expected ∆X. The material deposited on the edge of the wall has to maintain the shape with low distortion for correct multilayering. The shape of the molten material depends on whether the surface tension and viscosity of the molten material at the DED process temperature are sufficient to withstand the shear stress produced by the gravitational forces. In addition, the time during which gravity can shape the molten material depends on the cooling rate that is, in turn, defined by each strategy.
As with Strategy A, the cladding head was tilted, dilution was directed toward the core of the wall, and the laser beam had an elliptical projection. Thus, the temperature was lower, and dilution into the surface of the wall was shallower. These facts, combined with a wider melt pool and a decreased clad height, which respectively contributed to a broader surface area securing the clad and to a more stable shape, reduced edge distortion due to gravity, permitting smaller angles. These results with Strategy A and Strategy B differed from the results of Möller et al. [16] that never achieved angles smaller than 70 • with the same strategies, using the same parameters as in the vertical-wall case, without considering the evolution of the single-clad characteristics linked to head tilting and the maximum possible value of ∆X.
In contrast, a vertical DED process was employed for Strategy B. The DED process therefore presented no elliptical laser spot projections or changes to the geometry of the molten material. Additionally, as the laser beam was vertical, the temperature and the dilution on the surface of the wall increased, and the shape of the deposited material presented higher distortion due to gravity. This fact reduced the increment on the X axis (∆X) of each layer, and after several layers, the laser beam was directed toward an area with no additional material. As the minimum value of ∆X at 70 • employing Strategy B with both nozzles at 500 mm·min −1 was 0.3 mm, it was defined as a new requirement an increment ∆X lower than 0.3 mm with Strategy B. Thus, it was necessary to adapt the process parameters to this requirement. As the walls obtained at 900 mm·min −1 could meet this requirement, they were included in the study.
At 900 mm·min −1 , it was possible to obtain walls with the four-stream nozzle, but in the walls fabricated with the continuous coaxial one, each new layer presented higher distortion and lack of material in the edges of the wall, reducing the length of the wall progressively ( Figure 15). Walls of 20 mm in length were done at the same conditions of the study for showing the severity of this phenomenon in the shape of the wall (Figure 16). Walls at lower laser power (1900 W and employing the equations for the dimensions of single clads) were also done, to reject the melt-pool temperature as a possible cause, but all walls presented the same defect. The good results with the four-stream nozzle with the same strategy suggest the influence of the different powder flow characteristics (trajectory, velocity, powder concentration, etc.) of both nozzles on the melt-pool dynamics as a possible cause. Thus, to obtain walls without distortions with the continuous coaxial nozzle is necessary a ∆X < 0.15 mm per layer, as in the wall obtained at 80 • . This requirement sharply reduces the layer height to maintain the wall angle. Further studies with the continuous coaxial nozzle were rejected, because it forces to work at 70 • within a range with an efficiency of the single clad similar to the four-stream one, but with an MDR lower than 0.5 kg·h −1 and, at 60 • , out of the range of parameters used in this study, in a range more typical of 1 KW system, like the one used in the works of Bartkowski et al. [30,31].
surface of the wall was shallower. These facts, combined with a wider melt pool and a decreased clad height, which respectively contributed to a broader surface area securing the clad and to a more stable shape, reduced edge distortion due to gravity, permitting smaller angles. These results with Strategy A and Strategy B differed from the results of Möller et al. [16] that never achieved angles smaller than 70° with the same strategies, using the same parameters as in the vertical-wall case, without considering the evolution of the single-clad characteristics linked to head tilting and the maximum possible value of ΔX.
In contrast, a vertical DED process was employed for Strategy B. The DED process therefore presented no elliptical laser spot projections or changes to the geometry of the molten material. Additionally, as the laser beam was vertical, the temperature and the dilution on the surface of the wall increased, and the shape of the deposited material presented higher distortion due to gravity. This fact reduced the increment on the X axis (ΔX) of each layer, and after several layers, the laser beam was directed toward an area with no additional material. As the minimum value of ΔX at 70° employing Strategy B with both nozzles at 500 mm·min −1 was 0.3 mm, it was defined as a new requirement an increment ΔX lower than 0.3 mm with Strategy B. Thus, it was necessary to adapt the process parameters to this requirement. As the walls obtained at 900 mm·min −1 could meet this requirement, they were included in the study.
At 900 mm·min −1 , it was possible to obtain walls with the four-stream nozzle, but in the walls fabricated with the continuous coaxial one, each new layer presented higher distortion and lack of material in the edges of the wall, reducing the length of the wall progressively ( Figure 15). Walls of 20 mm in length were done at the same conditions of the study for showing the severity of this phenomenon in the shape of the wall (Figure 16). Walls at lower laser power (1900 W and employing the equations for the dimensions of single clads) were also done, to reject the melt-pool temperature as a possible cause, but all walls presented the same defect. The good results with the four-stream nozzle with the same strategy suggest the influence of the different powder flow characteristics (trajectory, velocity, powder concentration, etc.) of both nozzles on the melt-pool dynamics as a possible cause. Thus, to obtain walls without distortions with the continuous coaxial nozzle is necessary a ΔX < 0.15 mm per layer, as in the wall obtained at 80°. This requirement sharply reduces the layer height to maintain the wall angle. Further studies with the continuous coaxial nozzle were rejected, because it forces to work at 70° within a range with an efficiency of the single clad similar to the four-stream one, but with an MDR lower than 0.5 kg·h −1 and, at 60°, out of the range of parameters used in this study, in a range more typical of 1 KW system, like the one used in the works of Bartkowski et al. [30,31].  The theoretical W i , L w , and H w of the walls at 500 mm·min −1 were quite close to the measured values. This fact allows for good approximations of ∆X and ∆Z. On the contrary, the wall width (W w ) presented higher values at smaller angles, compared with the calculated ones, and a high waviness, due to wall edge distortion during the DED process. These effects must be considered in subsequent machining processes, as explained by Ostra et al. [32]. The results with Strategy B at 900 mm·min −1 presented the same trend as at 500 mm·min −1 . Despite the defects presented by the walls manufactured with the continuous coaxial nozzle, the measured characteristics of the part unaffected by the distortion were quite close to the theorical dimensions, except for W w, as in the other cases.
four-stream nozzle; (b) distortions in the edges of the walls; (c) cross-section at different conditions. The theoretical Wi, Lw, and Hw of the walls at 500 mm·min −1 were quite close to the measured values. This fact allows for good approximations of ΔX and ΔZ. On the contrary, the wall width (Ww) presented higher values at smaller angles, compared with the calculated ones, and a high waviness, due to wall edge distortion during the DED process. These effects must be considered in subsequent machining processes, as explained by Ostra et al. [32]. The results with Strategy B at 900 mm·min −1 presented the same trend as at 500 mm·min −1 . Despite the defects presented by the walls manufactured with the continuous coaxial nozzle, the measured characteristics of the part unaffected by the distortion were quite close to the theorical dimensions, except for Ww, as in the other cases. When employing Strategy A with the four-stream nozzle, the measured widths and heights presented a pattern in relation to the wall angle: both the measured dimensions of Wi and Ww increased when reducing the wall angle, while the measured values of Hw and Lw decreased, thereby producing wider and shorter walls with the same efficiency. This evolution should be considered when defining the manufacturing toolpaths.

Conclusions
The optimum conditions for processing Alloy 718 with DED in terms of quality (no defects) powder efficiency (maximum), MDR (>0.5 kg·h −1 ), and aspect ratio (>3 and <5) were a laser power of 2500 W, a feed rate of 500 mm·min −1 , and a powder mass flow of 18 and 14 g·min −1 for the four-stream nozzle and the continuous nozzle, respectively. The results obtained for these conditions were a powder efficiency of 60.1% and 85.4%, and an MDR of 0.65 and 0.71 kg·h −1 , for the four-stream and the continuous nozzle, respectively.
As regards the head tilting effect and non-perpendicular deposition with the four-stream nozzle, the clad characteristics showed no obvious effects of gravity except at 30° with Strategy 1 (α = β). At that angle, clad asymmetry at 500 mm·min −1 in the direction of gravity was higher than in other cases. This is because shear stress resulting of gravity increases with the tilting angle. Before the solidification, at 30° the viscosity is not enough to resist the shear stress. As each layer of the melt pool moves faster than the one just below, the distortion increases with the height. In contrast, Strategy 2 (α = 0) and Strategy 3 (β = 0) presented a widened width and a decreased height and area at a larger angle. Thus, the increases of the width and the reduction of the height reduced distortions of the melt pool caused by shear stress. In addition, the wetting angle decreased increasing the adhesive forces to the substrate.
The gravity effect of the continuous nozzle type strongly reduced the powder efficiency whenever strategies required a tilted head (α ≠ 0).
As a possible application, tilting the four-stream coaxial nozzle, as per Strategy 3 (β = 0), could be an interesting option where coatings of lower thickness are required. Contrary to what happens When employing Strategy A with the four-stream nozzle, the measured widths and heights presented a pattern in relation to the wall angle: both the measured dimensions of W i and W w increased when reducing the wall angle, while the measured values of H w and L w decreased, thereby producing wider and shorter walls with the same efficiency. This evolution should be considered when defining the manufacturing toolpaths.

Conclusions
The optimum conditions for processing Alloy 718 with DED in terms of quality (no defects) powder efficiency (maximum), MDR (>0.5 kg·h −1 ), and aspect ratio (>3 and <5) were a laser power of 2500 W, a feed rate of 500 mm·min −1 , and a powder mass flow of 18 and 14 g·min −1 for the four-stream nozzle and the continuous nozzle, respectively. The results obtained for these conditions were a powder efficiency of 60.1% and 85.4%, and an MDR of 0.65 and 0.71 kg·h −1 , for the four-stream and the continuous nozzle, respectively.
As regards the head tilting effect and non-perpendicular deposition with the four-stream nozzle, the clad characteristics showed no obvious effects of gravity except at 30 • with Strategy 1 (α = β). At that angle, clad asymmetry at 500 mm·min −1 in the direction of gravity was higher than in other cases. This is because shear stress resulting of gravity increases with the tilting angle. Before the solidification, at 30 • the viscosity is not enough to resist the shear stress. As each layer of the melt pool moves faster than the one just below, the distortion increases with the height. In contrast, Strategy 2 (α = 0) and Strategy 3 (β = 0) presented a widened width and a decreased height and area at a larger angle. Thus, the increases of the width and the reduction of the height reduced distortions of the melt pool caused by shear stress. In addition, the wetting angle decreased increasing the adhesive forces to the substrate.
The gravity effect of the continuous nozzle type strongly reduced the powder efficiency whenever strategies required a tilted head (α 0).
As a possible application, tilting the four-stream coaxial nozzle, as per Strategy 3 (β = 0), could be an interesting option where coatings of lower thickness are required. Contrary to what happens when increasing the feed rate, tilting the nozzle reduces the height, but also increases the width of the clad, with barely any powder efficiency loss. This change of geometry permits a wider range of coating thicknesses, changing only the overlap percentage without changing the DED process parameters.
The strategy and the type of nozzle define the maximum horizontal displacement per layer (∆X) that can be reached for obtaining inclined walls, without distortions. As the angle of the wall (θ) and ∆X define the layer height and, as a result, the amount of material deposited per layer, it is necessary to select the single clad geometry according to these parameters.
Although Strategy A results in the same powder efficiency and MDR for different wall angles, the wall geometry changes, increasing the waviness and generating wider and shorter walls at small wall angles.
In future developments, it would be advisable to address the following aspects: • To extend the range of angles in this study of the effects of head tilting and non-perpendicular deposition with all the strategies in the case of the four-stream nozzle and only with Strategy 2 in the case of the continuous nozzle.

•
To extend the range of angles for the manufacture of inclined walls with both strategies in the case of the four-stream nozzle, without exceeding an ∆X higher than 0.3 mm, employing Strategy B.

•
To develop geometric models for coatings and walls by using the Equations for the single-clad characteristics and the evolution of the characteristics linked to head tilting and non-perpendicularity.

•
To use the models that are developed to manufacture walls of variable thickness.

•
To include the models developed in Computer Aided Manufacturing software, in order to calculate the DED process toolpath, the Computer Assisted Design expected from the DED process, and the subsequent machining toolpath.

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