Figure 6 shows the calculated shrinkage cavity distribution positions in the simulation software for Case A. It can be seen that the main shrinkage cavities of the casting are located where the large nozzle wall and the small nozzle wall overlap, which is also where the thickness difference of the casting is greatest. The thickness from the large tube orifice to the small tube orifice is about 60mm, and the average thickness of the castings surrounding the large and small tube orifices, excluding the frame, is about 14mm. Such a large thickness difference causes delayed solidification of the large and small tube orifices, which in turn causes shrinkage defects to be distributed in the tubes around the mouth.
Figure 7 shows a comparison between the predicted A3 section of defects in Case A and the X-ray photo of the casting actually located on the same section. As shown in Fig. 7 (a) and Figure (b), the shrinkage cavity defects are all indexed in the thickness between the small tube opening and the large tube opening, and the distribution position is very close to the position detected by X-ray in Fig. 7 (c). The red circle in the X-ray photo is where shrinkage holes actually exist. Figure 7 (c) was reproduced from a traditional film. Since we cannot borrow a scanner that professionally scans X-ray films, capturing the outline of the defect on the film is difficult.
Figure 8 shows the X-ray results of the A7 section of other trial castings. Since A3 and A7 are symmetrical two sides of the casting geometry and are also the areas where the simulated shrinkage defects are mainly distributed, the actual X-ray inspection results of the shrinkage defects in the A3 and A7 sections are distributed in similar areas, that is, in this section the thickest area, the location where the defect occurs is the red circle in Fig. 8.
Figure 9 shows the expected defect occurrence areas generated by solidification shrinkage simulations for Cases A to C. The prominent locations where shrinkage defects occur are the same in the three Cases. This is a problem caused by the geometric characteristics of the casting itself. We can only slow down the occurrence of defects through the design of the tree. The defects of all solutions occur in the thickest parts of the large and small nozzles of the casting, and there will be no prediction of casting shrinkage in other areas.
Although the locations of defects are similar, it can be seen that the distribution of casting shrinkage on the left and right sides of the nozzle in Case B has been significantly reduced the shrinkage compared to Case A. This is because the area covered by the triangular insulation cotton slows down the solidification speed of the thin parts of the casting. Reduce the large-scale independent liquid phase zone inside the casting's large nozzle and small nozzle, thereby allowing the solidification trend to shrink toward the die head as continuously as possible.
As for the difference between Case C and Case B, the only difference is the four bracket arms. The four bracket arms make the wax discharge smoother and play the role of exhaust during the casting process. They can also be used as a structure to facilitate the clamping when the operator moves the tree. Most importantly, during the simulation, adding these four bracket arms also helped reduce shrinkage cavities in castings. Figure 9 shows Case C has fewer defects than Case B. The actual numbers will be described in detail later. The reduction of shrinkage holes should be because the extra material from the four bracket arms can provide some shrinkage. It also has a little thermal insulation effect on the gate, which makes it completely solidify later, allowing more soup from the die to replenish the casting part.
The casting is cut into nine pieces and numbered A1-A9. The simulation model is cut into nine equal parts according to the cross-section method of the actual sample delivery, as shown in Fig. 2 (a)-(c). The trial castings made in Case C do not have actual cross-sections and are sent for testing. Instead, the castings are irradiated from multiple angles to avoid any geometric features of the castings that would cause internal defects not to be detected by X-rays. However, the simulation of Case C is also cut into nine sections such as A1-A9 to predict the location of defects in order to compare whether the distribution position of the casting shrinkage is different from Case A.
As shown in Fig. 9(c), The casting defects in Case C will be mainly distributed on the cross-sections of A3 and A7 slices, while small areas of defects will still exist in A5. Compared with the simulation results of Case A in Fig. 9(a) and case B in Fig. 9 (b), the simulated defects on the A3 and A7 slices of Case C are significantly reduced.
Figure 10 shows the prediction of the exact percentage of elements containing shrinkage cavities in the castings from Case A to C. When the triangular insulation cotton area covered on the tree group increases, the percentage of elements containing shrinkage holes decreases (the PES value of plan B is nearly 41% less than that of Case A), which proves that the changed insulation cotton plan causes the internal solidification trend of the casting. It is relatively continuous, alleviating the problem of independent liquid phase zone generation when the thickness difference of castings is too large, allowing the die-feeding function to be fully exerted, which can greatly reduce the occurrence of internal shrinkage cavities. Compared with Case B, Case C adds four new arms for wax discharge, exhaust and convenient clamping in on-site operations. The PES simulation prediction can also achieve the effect of feeding and reducing the generation of shrinkage cavities. In case C, the PES value is reduced by 24.6% compared to case B and 55.5% compared to case A. The slight difference in a good tree assembly plan can significantly improve the quality of castings.
Figure 11 (a)-(c) shows the location of virtual thermodynamic sensors (VTDS) placed in each gate system to monitor the temperatures during operation [2] for Case A, Case B, and Case C, respectively. In the schematic, VTDS numbered 1 to 3 are marked on the red dots. The VTDS is placed at the die head's and gate's center, respectively. The position of the VTDS can clearly describe the solidification trend of the entire casting. The solidification time of the gate can be used better to compare the amount of shrinkage cavities inside the casting. The later the gate solidifies, the more molten steel located in the die head enters the cavity of the casting, which means that Fewer shrinkage defects will be generated.
At VTDS, Fig. 12 (a)–(c) illustrates the time function of the temperature change of the molten metal for three different casting plans: (a) Case A, (b) Case B, and (c) Case C. In Fig. 12(a) and Fig. 12(b), we observe that the molten iron change from its liquidus temperature to its solidus temperature after 2:50 seconds in Case A and 95 seconds in Case B respectively. This time difference shows that adjusting the insulation cotton covering method on the group tree can prevent the solidification process of metal flowing from the runner to the casting by nearly 4 to 5 seconds, nearly twice the time, and allow more molten metal to flow into the mold cavity of the casting, causing Fewer shrinkage voids.
According to Fig. 12(b) and Fig. 12(c), it can be seen that Case B requires 95 seconds and Case C requires 110 seconds for the molten iron to drop from the liquidus temperature (1400°C) to the solidus temperature (1375°C) at points 2 & 3. This time difference shows that fine-tuning the geometry of the casting system can prevent the solidification process of metal flowing from the runner to the casting by nearly 15 seconds and allow more molten metal to flow into the mold cavity in the casting area, resulting in fewer shrinkage porosity.
Figure 13 shows the inspection results after X-ray inspection of the trial casting produced using the tree assembly method and process parameters of the final plan C. This time, we made a trial run of five series trees, that is, there were ten valve castings available for inspection. Five of the ten trial castings were not found to have detectable internal shrinkage cavities, as shown in Fig. 13(a), (b), (e), (i) and (j). As for the other five test castings with detected shrinkage cavities, the locations of the shrinkage cavities are all distributed at the slice section A5, similar to the prediction in Fig. 13. As for the A3 and A7 cross-section locations with many internal shrinkage cavities in the initial plan, no shrinkage cavities were detected in these ten test castings. It can be confirmed that adjusting the triangular insulation cotton coverage area and fine-tuning the group tree geometry can significantly reduce the occurrence of shrinkage cavities—position on the casting. For the trial castings with shrinkage cavities, the sizes of the shrinkage cavities are as follows: There are two shrinkage cavities of 0.84mm and 0.94mm in Fig. 13(c), 1.80mm in Fig. 13(d), and 0.64mm in Fig. 13(f). One mm shrinkage hole, one 1.60mm shrinkage hole in Fig. 13(g), and one 2.49mm and 1.78mm shrinkage hole in Fig. 13(h). Special attention is paid to the fact that in the middle and outer parts of these ten inner castings, where the thickness is uniform, the inspector has circled some shadows in some places, suspecting the occurrence of sponge-like shrinkage cavities. However, from the perspective of simulation and theory, large-area sponge-like shrinkage cavities can't occur at this location, so a few of them were picked out for cross-sectional inspection. If sponge-like shrinkage cavities occur, the cross-sectional inspection can detect them. It can be concluded that the final result is that there is no sponge-like shrinkage cavity. Judging from the international standard of ASTM E2660, the internal shrinkage detection results of the ten test castings were all below level four, which met the customer's requirements.
In order to verify the accuracy of the simulation results, the valve parts were lost-wax cast in the foundry using the casting parameters derived from CAE X-ray inspection used to evaluate the effectiveness of design changes in reducing or avoiding casting defects. Figure 14(b) illustrates a photo of the wax mold set tree. Figure 14(c) shows the rough embryo of the assembled tree casting after the ceramic shell mold was shaken off after casting under Plan C. It is confirmed whether there are any defects on the surface of the casting. Figure 14(d) shows the X-ray image of the casting result, proving the effectiveness of the casting system design and process parameters in Case C. As shown in Fig. 14(d), in the X-ray photograph, no black spots were detected inside the valve body of Scheme C, meaning no internal defects were produced. Therefore, all the above results provide rigorous evidence for the casting quality of the trial cast stainless steel (316L) valve parts in Case C, improving the effectiveness of valve casting production scheme.