Contributors: Cao Fusheng, Lu Zhongliang
Contributed by: State Key Laboratory of Mechanical Manufacturing System Engineering
Source: Chinese Society of Mechanical Engineering Additive Manufacturing Technology (3D printing) Branch
In the study of thermal behavior, physical experiments are largely limited due to the limited range of controllable parameters, low efficiency, and difficulty in in-situ microscopic characterization. In contrast, numerical simulations in this area are of increasing interest to researchers and engineers. A lot of work has been done on the thermal behavior of monorail printing process. However, there will be obvious thermal effect between two adjacent tracks, which will lead to the change of the molten pool size of the track after printing. At the same time, in the LPBF additive manufacturing process, this phenomenon can also be observed between two adjacent layers. Therefore, a comprehensive understanding of the thermal behavior in the printing process is of great significance for effective control and accurate prediction of process-structure-performance. The following studies have numerically reproduced the multi-rail melting process of 316L powder in LPBF by computational fluid dynamics (CFD) method, which provides a valuable reference for obtaining high performance products with the required accuracy.
Figure 1 shows the multi-rail melting process of 316L stainless steel under typical conditions, where the top view of each moment is shown in Figure 1(a). It can be seen that at L0, the laser is just acting on the powder bed, and the temperature of the laser spot center is high and the temperature of the other regions is low follows a Gaussian distribution. During the scanning of the first orbit (L1), the temperature of the molten pool presents a comet-like distribution. Figure 1(b) shows the temperature contour of the molten pool cross-section at each moment. It can be found that the depression of the molten pool caused by the recoil pressure gradually deepens, which further shows the significant effect of the preheating of the already printed molten layer. In addition, the velocity vector of the local region of the melt pool in the 1st and 7th orbits is analyzed in the subgraph labeled "local". Compared with the first orbit, the velocity vector of the molten pool in the 7th orbit is larger, indicating that the recoil pressure in the molten pool also increases during the multi-orbit melting process, which confirms the above conclusion.
Figure 1 Multi-rail melting process: (a) Top view; (b) Cross section; (c) Temperature evolution of the probe line from moment to moment
The thermal influence Angle is used to quantify the thermal interaction between melt layers and is defined as the Angle between the line connecting the bottom of the first and last melt layers and the horizontal line. Figure 2 (a) shows the thermal influence Angle of the multi-track melt layer at different laser power, scanning speed, and hatch spacing. To make things more intuitive, the color and size of the circle are used to represent the value of the Angle. It can be found that when the laser power is 240 W, the scanning speed is 0.4 m/s, and the hatch spacing is 60 μm, the maximum thermal effect Angle of the multi-rail melt layer is 17.3°. Away from this point, the thermal influence Angle decreases. Figure 2 (b) depicts the depth of the multi-track melt layer under different conditions. It should be emphasized that the depth here represents the average distance (X middle) between the upper and lower boundaries of the melt layer and the different Y coordinates at the selected cross section. For the heat influence Angle, the color and size of the circle are used to indicate the depth of the melt layer. It can be seen that when P = 240 W, V= 0.4 m/s and H = 60 μm, the maximum depth of the multi-rail melt layer is 130.93 μm. Away from this position, the depth gradually decreases.
FIG. 2 Thermal influence Angle (a) of molten layer and depth of molten pool (b) under different conditions
The three-dimensional morphology (left picture) and corresponding cross sections (right picture) of the multi-orbit melt layer obtained from different conditions are shown in Figure 3. When other conditions are fixed, the thermal influence Angle of the multi-rail melt layer at P =240 W is significantly greater than that at P = 160 W, and the depth of the melt layer is also greater in the latter case. In addition, the melt layer fluctuation (standard deviation/mean value of molten pool height) under different P is calculated according to the cross section. The results show that the fluctuation of P =160 W (0.0176) is much lower than that of P = 240 W (0.0318). This can be attributed to the large pool size and strong Marangoni convection in the pool formed by large thermal energy at high laser power, resulting in more convexation of the melt layer. In addition, because each melt layer forms a large gully between the adjacent melt layers, the surface fluctuation of the multi-rail melt layer is more serious at high power.
Figures 3 (b1) and (b2) depict the 3D morphology and cross sections of the multi-orbit melt layer formed at different scanning speeds. A small scanning speed results in relatively high fluctuations in the melt layer. In addition, the interlayer overlap thickness of the multiorbital melt layer corresponding to the high scanning speed is very low. Although there are no obvious defects here, it is foreseeable that non-melting defects are prone to occur when the thickness of the powder bed increases slightly.
Figure 3 (c1) and (c2) show the three-dimensional morphology and cross-section of the multi-track melt layer at different hatch spacing H. The influence of hatch spacing on surface fluctuation is not regular. This is despite the fact that small hatch spacing produces large surface fluctuations, such as high laser power and low scanning speeds. Due to the large axial distance, the high hatch spacing also causes large dimensional fluctuations between the melt layers.
FIG. 3 Morphologies and corresponding cross sections of the molten layer formed under different conditions, where: (a1, a2) different laser power; (b1, b2) different scanning speeds; (C1, C2) Different hatch spacing
The BPNN model is established and adopted. Figure 4 (a) shows a comparison between the predicted and simulated heat effect angles under different conditions, and very good agreement can be determined. Both the predicted and simulated values are tightly distributed on or along Y=X, which means that the prediction accuracy is high. The BPNN model can well predict the depth of molten pool under different conditions. Therefore, the results in Figure 4 show that the current BPNN model can be successfully used to predict the thermal influence Angle and depth of the multi-orbit melt layer under any combination conditions. In other words, by using the BPNN model, the thermal control chart can become more accurate.
FIG. 4 Numerical simulation and BPNN prediction of heat influence Angle (a) and molten pool depth (c) under different conditions; Correlation between heat influence Angle (b) and melt pool depth (d) in numerical simulation and BPNN prediction
References:
Yao D, Wang J, Luo H, et al. Thermal behavior and control during multi-track laser powder bed fusion of 316 L stainless steel[J]. Additive Manufacturing, 2023, 70: 103562.