In the process of LAM, drastic change in temperature is the main cause of the residual stress. It is important to observe the temperature changes currently to analyze the residual stress conveniently later. As the processing progresses, the temperature of the heat source of the LAM increases (see Fig. 6).
Residual stress is our focus in simulating LAM. The power of the laser heat source is one of the most important reasons for the residual stress. Therefore, simulation models for different powers are established to verify the distribution law of residual stress under different machining powers, as shown in the Fig. 7. Before that, we set up several reference points to detect the residual stress around. As mentioned above, two layers of each material were printed, and the middle common nodes of the two cladding layers was used as the reference points. It can be seen from the Fig. 7 that with the increase of laser power, the residual stress gradually increases, and the two are positively correlated. However, due to the influence of the material gradient in the stacking direction, the influence of different cladding layers is also different. The laser power received at the bottom layer of the structure (100% Inconel718) has a greater influence, while the upper layer of the structure, especially the layers 7 and 8(20% Inconel718 and 80% TC4), the influence of the laser power is smaller. The residual stress distribution of the overall structure presents an “inverted bowl shape”, the residual stress on both sides is small, and a large residual stress is generated in the middle part of the structure. In actual engineering, we should pay more attention to the middle part of the structure, and the fracture phenomenon of a large number of additively manufactured specimens occurs mostly in the middle of the structure.
Compared with isotropic materials, the biggest feature of this paper is the material anisotropy of the structure in the stacking direction. Due to the differences in the physical and mechanical properties of materials, materials with different mixing ratios have different resistance to residual stress. In order to facilitate the experimental preparation of functionally graded materials, simulation models with different transition ratios are designed, which are divided into four types: 6:4, 7:3, 8:2 and 9:1. The LAM simulation model designed in this paper has a total of 10 cladding layers, of which the 1st and 2nd layers are 100% Inconel718 material, the 9th and 10th layers are 100% TC4 material, and the 5th and 6th layers are 50% TC4 and 50%Inconel718 material mixing, these three parts always remain unchanged, the 3rd, 4th, 7th and 8th layers are designed as transition layers, taking the transition ratio of 8:2 as an example, that is, the 3rd and 4th layers are 80%Inconel718 mixed with 20% TC4 material, the 7th and 8th layers are mixed with 20% Inconel718 and 80% TC4 material. As mentioned above, the residual stress mainly takes a large value in the middle part of the structure. In order to facilitate the detection of residual stress, a total of 15 reference points is averagely set on three lines, which are called "left line", “middle line” and “right line” (see Fig. 8(d)). These three lines are taken at the quarter points of the scanning direction of the structure, that is, the two adjacent lines are separated by 45 meshes (1.5cm). Figure 8(a), (b) and (c) show the residual stress distribution on these three lines. Due to the influence of the gradient, the variation of the residual stress does not have an obvious regularity like the laser power. In general, when the composition ratio of the transition layer is 6:4 and 9:1, the residual stress of the structure changes most obviously, which is obviously not conducive to the stability of the structure. When the composition ratio of the transition layer is 7:3 and 8:2, the change of residual stress is relatively moderate, and when the two are compared, the composition ratio of 8:2 has a smoother residual stress transition, which is not easy to cause the buckling of the structure. and breakage issues. Therefore, when using the LAM method to prepare the TC4/Inconel718 functionally graded material, it is more in line with the actual needs to use the transition layer ratio of 8:2.
In the end, in order to reflect the difference between the PAE method and traditional “Model Change” method, the calculation time of the above models are listed, as shown in Table 4. Compared with the “Model Change” method, the PAE method has relatively high computational efficiency and can save a lot of computational time. When calculating anisotropic materials, “Model Change” method will consume a lot of time for stress analysis due to the change of the material composition of the transition layers. This advantage of PAE will be more obvious when calculating gradient materials. When calculating isotropic materials (such as TC4 alloy), the calculation efficiency can be increased by 995%, and when calculating anisotropic materials, the higher the laser power, the more obvious the efficiency improvement of the PAE method; the calculation time of combining these models, the PAE method can probably improve by about 1650% than the “Model Change” method. It is worth stating that the CPU used in this paper is Intel® Core™ i5-10600KF CPU @ 4.10GHz, and the utilization rate is kept above 90%. The memories are two Essencore DDR4-3200 8GB and two Crucial Technology DDR4-2666 8GB.
Table 4
Calculation time of above models
|
Model
|
Calculation time
|
Efficiency
|
PAE
|
Model Change
|
Isotropic
|
Pure TC4
|
3h15mins
|
35h44mins
|
995%
|
Anisotropy
|
400W
|
3h30mins
|
71h18mins
|
1937%
|
500W
|
4h15mins
|
92h36mins
|
2078%
|
600W
|
5h3mins
|
108h36mins
|
2050%
|
6:4
|
4h47mins
|
113h7mins
|
2265%
|
7:3
|
4h10mins
|
98h25mins
|
2262%
|
8:2
|
4h15mins
|
94h36mins
|
2126%
|
9:1
|
4h39mins
|
116h14mins
|
2400%
|