I have been engaged in the research and production of tractor components for many years. The divided axle housing is one of the most critical load-bearing parts of a tractor. In practice, the casting quality of the axle housing directly affects the service life and safety of the whole vehicle. However, during the sand casting process, defects such as shrinkage cavities and porosity often occur, which are typical sand casting defects. According to statistics, the occurrence rate of these sand casting defects in the divided axle housing was as high as 70%, leading to oil leakage in service and causing annual economic losses of more than one million RMB. To solve this problem, I conducted a systematic study based on the actual production conditions in our foundry. I designed three different gating systems, used the ProCAST finite element software to simulate the mold filling and solidification processes, and optimized the casting design by adding risers and chills. Finally, I verified the optimized process through cutting inspections and air tightness tests. The results show that the new semi-closed gating system can effectively reduce sand casting defects, and the rejection rate dropped from 70% to about 3%. This paper presents the complete research process and the main conclusions.
1. Casting Structure Analysis and Process Parameter Determination
The divided axle housing in this study is made of ZG40Mn low-alloy cast steel. Its main external dimensions are 452 mm × 340 mm × 396 mm, with an average wall thickness of 25 mm. The maximum wall thickness is 29 mm and the minimum is 15 mm. The axle housing must be free of shrinkage cavities, shrinkage porosity, cracks, cold shuts and other sand casting defects. The material composition and mechanical properties are listed in Tables 1 and 2.
$$ \text{Table 1: Chemical composition of ZG40Mn (wt.%)} $$
| Material | C | Mn | Si | S | P |
|---|---|---|---|---|---|
| ZG40Mn | 0.35–0.45 | 1.0–1.5 | 0.35–0.45 | ≤0.03 | ≤0.03 |
$$ \text{Table 2: Mechanical properties of ZG40Mn after normalizing and tempering} $$
| σs | σb | δ5 | Ψ | Hardness |
|---|---|---|---|---|
| ≥295 MPa | ≥640 MPa | ≥12% | ≥30% | 163–200 HB |
For sand casting of steel, the quality of raw materials and auxiliary materials is very important. The raw materials include low-carbon steel scrap, ferrosilicon, high-carbon ferromanganese, and aluminum-iron alloy deoxidizer. The silica sand must meet the requirements of GB/T 9442-1998. I strictly controlled the composition and cleanliness of all materials to minimize the possibility of sand casting defects caused by inclusions and gas generation.
The pouring position was determined by considering the critical surfaces of the axle housing. The flange face and the inner cylindrical cavity are the most important functional areas. I chose a horizontal pouring position with the parting plane in the middle, which is beneficial for smooth filling and easy mold making. The final parting surface is shown in Figure 1.
The pouring temperature was chosen based on simulation of the local temperature curves for different pouring temperatures. I compared 1540 °C, 1560 °C and 1580 °C. The results showed that 1580 °C gives relatively slow solidification in the later stage, which helps to avoid hot tears and reduce stress concentration. Therefore, the pouring temperature was set to 1580 °C. The casting shrinkage coefficient was taken as 2.0% according to the material and the structure. The dimension tolerance was chosen as CT12 level. The pattern draft angle was selected as 3° for the reinforcing rib area.
The sand core design was carried out for the internal cavity. The core has two horizontal core prints, each 50 mm long, with a draft angle of 3°. The vertical core print is 50 mm long at the top and 30 mm at the bottom, with draft angles of 4° and 1°, respectively. The core was made of resin-coated sand with good strength and collapsibility. The original casting process used a bottom gating system, which caused poor filling and severe shrinkage defects at the inner oil seal seat, as shown in Figure 2.
Based on the structure analysis, I identified that the original process had the following main problems: (1) the bottom gating system resulted in insufficient filling ability; (2) the long pouring time of 25 s caused premature solidification; (3) the lack of effective risers and chills created isolated liquid regions. Therefore, I decided to redesign the gating system completely.
2. Theoretical Analysis and Design of the Gating System
The gating system must ensure that the molten steel fills the cavity smoothly and quickly without causing turbulence, oxidation, or sand erosion. Since the original bottom gating system produced many sand casting defects, I selected a middle injection system. The inner gate location was set at the parting line, which is a compromise between top and bottom injection.
The minimum static pressure head was calculated using the following equation:
$$ h_M = L \tan \alpha $$
where \(L\) is the flow distance from the sprue center to the highest point of the casting, and \(\alpha\) is the pressure angle. In this design, \(L = 320\) mm and \(\tan \alpha = 0.176\), so \(h_M = 56\) mm. Thus the minimum required height of the sprue was 246 mm.
I used the pouring rate method to calculate the choke area. The total weight of the casting including gating system was about 108 kg. The average density factor was:
$$ \rho = \frac{G_L}{V_C} = \frac{108}{45 \times 34 \times 40} = 0.002 \ \mathrm{kg/cm^3} $$
From the empirical table, the coefficient \(C = 0.8\) and the pouring rate \(k = 0.6\ \mathrm{kg/(cm^2 \cdot s)}\). The pouring time was:
$$ t = C \sqrt{G_L} = 0.8 \times \sqrt{108} \approx 9.6 \ \mathrm{s} $$
I took \(t = 10\ \mathrm{s}\). The choke area \(A_{\mathrm{choke}}\) was calculated as:
$$ A_{\mathrm{choke}} = \frac{G_L}{k\, t\, S’} = \frac{108}{0.6 \times 10 \times 0.8} = 22.5 \ \mathrm{cm^2} $$
Here \(S’\) is the flow coefficient for cast steel, taken as 0.8. The choke area was the cross-section of the ingate. For a cylindrical ingate, the diameter was:
$$ d = 2 \sqrt{\frac{A}{\pi}} = 2 \sqrt{\frac{22.5}{3.14}} \approx 53.5 \ \mathrm{mm} $$
I adopted a diameter of 54 mm for the ingate.
According to the different cross-section area ratios of the gating system components, I designed three schemes:
- Scheme A (closed gating system): \( \sum A_{\mathrm{in}} : \sum A_{\mathrm{run}} : \sum A_{\mathrm{spr}} = 1 : 1.3 : 1.5 \)
- Scheme B (open gating system): \( \sum A_{\mathrm{in}} : \sum A_{\mathrm{run}} : \sum A_{\mathrm{spr}} = 1.2 : 1.1 : 1 \)
- Scheme C (semi-closed gating system): \( \sum A_{\mathrm{in}} : \sum A_{\mathrm{run}} : \sum A_{\mathrm{spr}} = 1 : 0.8 : 1.2 \)
All three schemes used one ingate with a cross-section of 22.5 cm². The cross-sections of the runner and sprue were determined accordingly. Table 3 summarizes the dimensions of the three gating systems.
$$ \text{Table 3: Cross-section areas of the three gating system schemes} $$
| Scheme | Ingate area (cm²) | Runner area (cm²) | Sprue area (cm²) | Area ratio |
|---|---|---|---|---|
| A: closed | 22.5 | 29.25 | 33.75 | 1 : 1.3 : 1.5 |
| B: open | 22.5 | 21.0 | 18.8 | 1.2 : 1.1 : 1 |
| C: semi-closed | 22.5 | 18.0 | 27.0 | 1 : 0.8 : 1.2 |
A ceramic foam filter was placed at the bottom of the sprue to trap slag and regulate the flow. The filter area was 30 cm².
3. Numerical Simulation of Mold Filling and Solidification
I used ProCAST to simulate the three gating systems. The three-dimensional models were built in Creo and imported into the Visual-Mesh module. The mesh size was set to 6 mm for the casting and 15 mm for the gating system. After surface and volume mesh generation, I checked the mesh quality and then set the initial conditions: pouring temperature 1580 °C, initial mold temperature 30 °C, interface heat transfer coefficient 500 W/(m²·K), and gravity acceleration 9.8 m/s². The material properties of ZG40Mn were obtained from the ProCAST database.
3.1 Solidification and Heat Transfer Equations
The numerical simulation of the solidification process is based on the heat conduction equation:
$$ \rho C \frac{\partial T}{\partial t} = \lambda \left( \frac{\partial^2 T}{\partial x^2} + \frac{\partial^2 T}{\partial y^2} + \frac{\partial^2 T}{\partial z^2} \right) + w $$
where \(T\) is temperature, \(t\) is time, \(\rho\) is density, \(C\) is specific heat, \(\lambda\) is thermal conductivity, and \(w\) represents the latent heat release rate during solidification.
3.2 Filling Process Analysis
Figure 3 shows the velocity field of Scheme A (closed gating system) at different filling stages. I observed that the molten metal flowed relatively smoothly, but there was slight splashing in the early stage. Scheme B (open gating system) showed many isolated liquid droplets during filling, which may cause cold shuts and oxide inclusions. Scheme C (semi-closed gating system) produced the most uniform and stable filling. The filling time was about 10 s for all schemes.
$$ \text{Table 4: Solid fraction (%) at different filling stages for the three schemes} $$
| Scheme | 25% filling | 30% filling | 64% filling | 96% filling |
|---|---|---|---|---|
| A: closed | 0 | 0 | 2.5 | 10.4 |
| B: open | 0 | 0.1 | 10.5 | 18.8 |
| C: semi-closed | 0 | 0 | 4.0 | 12.7 |
From Table 4, Scheme B had the highest solid fraction during filling, meaning the metal lost fluidity too early, which easily leads to cold shuts and misruns. Scheme A was acceptable, but Scheme C showed the best combination of smoothness and low solidification during filling.
3.3 Temperature Field and Solidification Analysis
The temperature field simulation allowed me to identify isolated liquid regions where shrinkage defects would form. For all three schemes, isolated liquid regions appeared in the later stage of solidification. Scheme B had the largest isolated liquid regions, and simulation predicted severe shrinkage porosity. Scheme A also had local isolated regions at the flange connection and the upper protrusions. Scheme C had smaller isolated regions, but shrinkage defects were still present inside the casting near the inner semi-axle oil seal seat.
$$ \text{Table 5: Solid fraction (%) at different times for the three schemes} $$
| Scheme | t = 8–12 s | t = 16–31 s | t = 71–82 s | t = 117–122 s |
|---|---|---|---|---|
| A: closed | 16 | 26 | 74.4 | 86.6 |
| B: open | 30 | 50.7 | 78.9 | 92.6 |
| C: semi-closed | 24 | 69.9 | 77.3 | 82.7 |
Scheme B obviously produced the worst solidification pattern. Thus, I abandoned Scheme B and continued optimizing Scheme A and Scheme C by adding risers and chills.
4. Design of Risers and Chills
To eliminate the shrinkage defects, I employed risers to provide additional molten metal during solidification, and chills to accelerate cooling in hot spots. The casting modulus was calculated as:
$$ M = \frac{V}{A} $$
For the critical section of the axle housing, the modulus was about 9.7 mm. I designed a riser with a diameter \(D_r = 80\) mm and height \(H_r = 150\) mm. The riser modulus was:
$$ M_r = \frac{D_r}{4} = 20 \ \mathrm{mm} $$
Since \(M_r > M\), the riser can provide effective feeding. I placed one blind riser at the ingate area, and one open riser on the upper surface of the casting. Additionally, I placed six chill blocks inside the inner cavity and one chill on the side wall. The chills were made of cast steel with dimensions of 30 mm × 50 mm × 60 mm, which create a strong chilling effect and promote directional solidification.
After modifying the three-dimensional model with risers and chills, I re-ran the ProCAST simulation. The results for Scheme A still showed some residual shrinkage defects near the inner flange. However, for Scheme C, the shrinkage defects were completely eliminated, and the porosity moved into the risers. This means the risers and chills worked effectively to feed the hot spots.
$$ \text{Table 6: Comparison of simulation results before and after adding risers and chills} $$
| Scheme | Original defects | After adding risers and chills |
|---|---|---|
| A: closed | Local shrinkage at flange and top | Defects reduced but not fully eliminated |
| C: semi-closed | Local shrinkage at inner oil seal area | No visible shrinkage defects |
Consequently, I selected Scheme C (semi-closed gating system) as the final optimized casting process. The final process parameters are summarized in Table 7.
$$ \text{Table 7: Final optimized casting process parameters} $$
| Parameter | Value |
|---|---|
| Gating system type | Semi-closed |
| Ingate area | 22.5 cm² |
| Runner area | 18 cm² |
| Sprue area | 27 cm² |
| Area ratio | 1 : 0.8 : 1.2 |
| Pouring temperature | 1580 °C |
| Pouring time | 10 s |
| Riser arrangement | Blind riser at ingate, open riser at top |
| Chill arrangement | Six chills in inner cavity, one on side |
5. Experimental Verification
To validate the optimized process, I produced prototypes using the semi-closed gating system. I first cast two pieces, then eight pieces in the second batch. All ten pieces were sliced at intervals of 10 cm. The cut surfaces were examined with a 10× magnifying glass. No visible cracks or pores larger than 3 mm were found. Later, 630 castings were produced in five batches, with 31 pieces randomly selected for destructive cutting. No internal defects were observed.
Next, I carried out air tightness tests on 20 machined axle housings from two batches. The test method involved sealing all openings with flanges and rubber gaskets, submerging the workpiece in water, and pressurizing the internal cavity to 0.5 MPa for at least 10 minutes. I checked for air bubbles to detect leakage paths caused by sand casting defects. Table 8 shows the test results.
$$ \text{Table 8: Air tightness test results for the optimized process} $$
| Test piece | Pressure (MPa) | Time (min) | Leakage | Result |
|---|---|---|---|---|
| A1–A8 | 0.5 | 10 | No | Pass |
| B1–B12 | 0.5 | 10 | No | Pass |
All 20 samples passed the air tightness test without any leakage. This indicates that the optimized sand casting process effectively eliminated the shrinkage porosity that used to cause oil seepage. By the end of 2019, more than 6000 axle housings had been produced using the new process. Only 174 were found to be defective, giving a rejection rate of approximately 3%, compared with the previous 70%.
6. Conclusion
In this work, I have successfully reduced sand casting defects in the tractor divided axle housing through systematic analysis, simulation, and experimental verification. The main conclusions can be summarized as follows:
- The middle injection gating system with a semi-closed design (ingate:runner:sprue = 1:0.8:1.2) provides the most stable mold filling and the best resistance to slag entrapment. It significantly reduces the risk of sand casting defects such as sand holes and inclusions.
- The pouring temperature of 1580 °C and pouring time of 10 s are optimal for this ZG40Mn cast steel axle housing.
- The use of combined risers and chills is essential to achieve directional solidification. The semi-closed process, together with one blind riser, one open riser, and six chills, completely eliminated shrinkage porosity in the critical areas.
- The air tightness test confirmed that the optimized process greatly improves the quality and reliability of the axle housing. The rejection rate due to sand casting defects dropped dramatically from 70% to around 3%.
This research demonstrates that computer simulation is a powerful tool for optimizing sand casting processes and reducing sand casting defects. In the future, I plan to extend this work to the coupled simulation of flow, temperature, and stress fields, and to use experimentally measured thermo-physical parameters to further improve the accuracy of defect prediction.