In this continuous coaxial nozzle system, powder is injected through an injection system with laser irradiation at the center of the nozzle, and two flow paths are employed. Figure 3 shows the 3D CAD model of the coaxial nozzle. As shown in Figure 4 the powder and the carrier gas are injected from the inner flow path, and the shielding gas is supplied through the outer flow path. The coaxial nozzle, having the powder feeding from the sideways and laser directed from the center of the nozzle, coincides with the powder at a focal point. From the side, there is a provision made for the entry of powder along with the carrier gas and shielding gas injection made through the pipe from the hopper. The respective dimensions of the coaxial nozzle is taken from BEAM machines (IREPA LASER, (Zhao et al., 2020). The nozzle is constructed in compartments of three. The first compartment is used for the laser beam projection. The second compartment is used for the travel of particles and carrier gas, for which eight inlets are given along the circumference of the nozzle. The last compartment, which is below the particle compartment, is for the shielding gas. The central and outer gas streams ensure the optics protection and oxidation protection, respectively. Normally, inert gas Ar or He is used as carrier and shielding gas, respectively. Hence, in this study, the effect of both gases on particle participation was analyzed numerically.

In this study, ANSYS Fluent software was used to simulate the 316L stainless steel powder stream characteristics. The stainless steel powder numerical simulations conditions were reported in Table 1. The gas phase was computed by the standard k- ε turbulent flow model that was adopted, and the powder stream was coupled as a discrete phase in Euler–Lagrange model, which has been already proved to be an effective method in previous similar studies. The characteristics of turbulent jet flow are momentum, heat and mass, which are transferred through flow at rates much greater than the laminar flow conditions. In such model, the governing equations for laminar flows are modified using the time-averaging method known as Reynolds averaging (Zhu et al., 2011; Morville et al., 2012; Arrizubieta et al., 2014). Addition of particle phase into the gas phase may integrally lead to a difference of mass, energy, and momentum. In this study, a Navier–Stokes equation-based computational model is used.

The governing continuity and momentum equations are expressed as Eqs 1, 2. ∇ ( ρ u ) = 0 (1) ∂ ∂ t ( ρ u ) + ∇ ( ρ uu ) = − ∇ p + ∇ ⋅ [ μ ( ∇ u + ∇ u T ) ] + ρ g (2) where ρ and μ are the mixture density and viscosity, respectively, p is the mixture mean pressure, u is the velocity of the mixture, and g is the gravitational acceleration.

The sediment particles are tracked in the Lagrangian reference frame to obtain their positions and velocities using the discrete particle model. As the particles are released from the nozzle, they are confined in the flowing inert gas. It plays a dominant role in determining the particle trajectories. Thus, the forces from the air phase are only used to calculate the particle’s movements in the present work as shown in Eq. 3. m p d u p d t = F D + F P + F V M (3) where m _p is the particle mass, u p is the particle velocity, F D is the drag force, F P is the pressure gradient force, and F V M is the virtual mass force. The definitions and calculation methods of these forces were presented in detail in the literature (Zhu et al., 2011).

The other additive equations used to describe the turbulent velocity fluctuation are as follows: the standard turbulence k- ε model put forward by Launder and Spalding (Takemura et al., 2019) is adopted; k equation and ε equation are expressed following the conservation of kinetic energy of turbulence: d d x i ( ρ ε u i ) = d d x i ( μ t σ k d k d x i ) + G k + G b − ρ ε (4) d d x i ( ρ ε u i ) = d d x i ( μ t σ ε d ε d x i ) + C 1 ε ε k ( G k + C 3 ε G b ) − C 2 ε ρ ε 2 k (5) G k = μ t ( d u j d x i + d u i d x j ) d u i d x j (6) G b = − g i μ t ρ P r t d ρ d x i (7) where G _k is the rate of production of turbulent kinetic energy due to average velocity gradient, G _b is the generation of turbulence owing to buoyancy, Pr_t is the turbulence Pradntl number, and σ k = 0.1 ,   σ ε = 1.3 ,   C 1 ε = 1.44 ,   C 2 ε = 1.92   a n d   C μ = 0.99 are empirical constants. The constant C 3 ε is equal to 1 when the direction of the gas stream is approximately parallel to the direction of gravity. The empirical constants are taken based on the published data (Li et al., 2021).

Following are the assumptions made in the design and simulation process of the work (Takemura et al., 2019):

• The 3D model of the nozzle has been designed in solid works, but actual simulation was carried out considering a 2D plane of the designed CAD model.

The CAD model is imported to Ansys, the plane of the substrate surface is subdivided into three planes in order to have provision to collect the particles which are being melted and the particles which are escaping from the melting zone. An imaginary plane is created between the nozzle exit and the substrate surface for the particles to concentrate at a point.

The 2D model is imported to the meshing tool where meshing and required boundary conditions are applied. E, designated as inlet 1 in Figure 5, is the inlet for powder particles and carrier gas. A, being inlet 2, is the inlet for shielding gas. B, representing the bin, is the boundary to collect melted particles. C, as the imaginary space bottom plane, is the boundary to collect unmelted particles. D is the nozzle wall boundary condition, and F specifies the space wall boundary conditions.

Meshing of element size 0.5 mm is applied to the geometry with quadrilateral element order as shown in Figure 6. A fine element size is considered in smoothing the mesh. All the remaining parameters of the mesh tool are taken as default parameters.

In simulation of the work, three models were considered. They are multiphase Eulerian model, k-ε turbulent model, and dense discrete phase model. In multiphase Eulerian model, two Eulerian phases are taken into consideration and one discrete phase is considered. In injection type, the surface injection with release of particles is selected from inlet 1 being the inlet for particles and carrier gas. The material of the particles is stainless steel 316L particles with Rosin–Rammler diameter distribution. The inlet velocity of the particles is taken as 0.25 m/s, and the total flow rate is 8 g/min. The particles are relatively spherical through the simulation process for every particle size.

To analyze the influence of Ar and He gas flow and the powder particle size on powder participation in the melt pool, a gas solid multiphase flow simulation model is conducted by using Ansys Fluent. The powder distribution analysis was made by assuming the baseplate under the powder nozzle. From the conclusions of Takemura et al. (2019), 2 to 3 L/min of carrier and shielding gas was found to have maximum powder efficiency for both numerical and experimental analyses. Hence, in this study, 2.5 L/min gas flow rate was considered for the numerical simulation.

To validate the numerically simulated results, grid indecency test was conducted by varying the number of mesh and number of cells of the nozzle design. The number of powder particles trapped in the bin was considered for all the mesh size with Ar as the carrier and shielding gas. It seems that, by increasing or decreasing the number of cells, the powder particles trapped in the bin are more or less the same, as depicted in Figure 7. Hence, 0.5-mm mesh size is considered for all the powder particle size distributions.

To verify the trajectory of powder flow and powder particle participation in the melt pool, the numerical simulation results were correlated experimentally using DED (IREPA) process. The experiments were made at a focal distance (Z) of 10 mm using 45–90 µm SS 316L powder, and the powder particles were collected and measured for the number of particles that participated and escaped from the zone. It was observed that the particle participation is almost 45% in the experimental condition as per the weight fraction measurement and is around 41.66% in the numerical simulation as can be seen in Figure 8. It has also been verified by depositing a single bead on the steel substrate. It was confirmed that the melting efficiency was about 51% as per the cross-section area of the bead (Figure 9).

To keep the experimental parameters similar as in the numerical computation, a single bead deposition was made as shown in Figure 7. The standoff distance was set to 10 mm from the nozzle outlet to the coaxial convergence of laser and powder in the flight mode. The process parameters of the laser cladding experiment are shown in Table 2. The length of the single-bead cladding layer was 80 mm. A cross-section of the single-bead cladding layer was obtained as shown in Figure 8, where the area of melted powder is 0.784 mm². To measure the area, Olympus optical profilometer was used.

The stainless steel powder particle flowability, convergence, and participation of powder particle in the melt pool trap were identified using flow 3D for several sizes. As shown in Figure 13 the particles converged at the focal point, which is formed at a distance of 12.59 mm from the nozzle exit and the width of powder distribution being 2.47 mm at the focal point. Figure 12(b) illustrates the number of particle participation in the melt pool trap, those that escaped, and those in the nozzle for both Ar and He as carrier gases for stainless steel powder. The particle simulation report was obtained from flow 3D for Ar and He as carrier gas.

For argon as carrier gas, the number of particles injected from the nozzle inlet was 600,000. The particles collected at the focal point and participating in melting are called trapped particles, the number of which was 505,245, and the number of particles that escaped was 80,573. At a particular time of trapping, there are 14,182 particles in the fluid.

For helium as carrier gas, the same number of particles was carried in He gas for the injection. The particles collected at the focal point, the number of which was 494,454, and the number of particles that escaped was 91,364. At a particular time of trapping, there are 14,182 particles in the fluid.

Given 60–70-µm powder particle size, Figure 14 shows the convergence for the 60–70 µm-range of particle size at a focal point which is formed at a distance of 11.86 mm from the nozzle exit. The width of powder distribution at the focal point was 4.48 mm. Figure 14 illustrates the number of particles participating in the melt pool trap, those that escaped, and those in the nozzle for both Ar and He as carrier gases for stainless steel powder. A particle simulation report was obtained from flow 3D of Ar and He gases. Results are shown considering a range of particle size of 60–70 µm, with argon gas as carrier and helium as shielding gas.

Similarly, simulations of powder particle sizes of 60–70, 78–80, 80–90, 90–100, and 45–90 µm were also done to verify the focal point of convergence and the width of powder convergence for both Ar and He as the carrier gases. Table 3 represents the particle convergence focal point distance, width, as well as powder particle participation in the melt pool.

Figure 15 illustrates the comparison between effects of particle size, focal distance, and the number of particles trapped in the melt pool with an assumption of bin below the focal point. It indicates that, for Ar gas, the particle trapping rate is high compared to He as carrier gas for any size of the particles.

Even though a greater number of particles was trapped at 50–60-µm particle size, from Table 3, it can be observed that the distance between the focal point and the nozzle exit was 12.59 mm, which was not optimum distance according to the experimental study conducted by various authors (Zekovic et al., 2007; Alvarez et al., 2018). The optimum distance between the focal point and the nozzle exit is around 12 mm, which is obtained for a particle size of 60–70 µm. The powder trapping efficiency was calculated as follows: P o w d e r   e f f i c i e n c y = M 2 M 1 where M2 = number of particles trapped and M1 = total number of particles injected.

Figure 16 shows the powder trapping efficiency of various powder size distributions for both Ar and He as carrier gas. It was observed that Ar as carrier gas has higher powder participation efficiency for all the powder size distributions, and 50–60 µm has the highest powder participation efficiency compared to the remaining size, but the focal point distance is around 12.6 mm which is close to that in practical experimental conditions that have the highest melting efficiency and uniform weld bead formation in laser direct energy deposition of 316 stainless steel powder (Wu et al., 2018). To meet the high tolerance standards required in laser direct energy deposition process, argon and helium are commonly used to provide inert atmospheres and protect the beads from the oxidation. Even nitrogen gas is used for carbon steel fabrication and other materials. The choice of optimal process gas depends on the material quality requirements. Even during the atomization process, argon is highly used for powder production, and also it will influence the microstructure. Hence, attempt is made here to verify the influence of gas on the powder particle as well as on the shielding effect. The shielding gas compacts the powder flow, reducing the effect of both the gravity force and carrier gas inertia that are mainly responsible for powder spreading in powder flow.

Powder flowability mainly depends on the distribution, size, and shape of powder particles. Sphericity is a significant factor for good flowability of metal powders. Particle size distribution is another important parameter—for example, a wide particle size distribution can affect the packing behaviors and, consequently, bead shrinkage and densification of molding parts.

The flowability of metal powders is not an inherent property—it depends not only on the physical properties (shape, particle size, humidity, etc.) but also on the stress state, the equipment used, and the handling method. The powder flow properties in different additive technologies are a complex area for study. Powder companies would like to avoid flow problems such as segregation, vaulting, and agglomeration and want to predict how a particular metal powder will flow and form/not form a homogeneous layer or compare the flow characteristics of metal powders with each other. Due to the expensive cost of metal powders, only a limited amount is provided for testing, so it is favorable to test different powder size distribution effects in DED process.