I conducted a comprehensive study on the sand casting process of a tractor divided axle housing, focusing on the elimination of shrinkage porosity and shrinkage cavity defects. The research combines theoretical analysis, numerical simulation using ProCAST finite element software, and experimental verification. Three types of gating systems were designed and compared: closed, semi-closed, and open. Through detailed simulations of mold filling and solidification, I determined that the semi-closed gating system with optimized riser and chill designs provides the best casting quality. The optimized process reduced the defect rate from 70% to approximately 3% in actual production. This paper elaborates on the entire process, including structural analysis, material selection, gating system design, numerical simulation, and experimental validation, with special emphasis on the role of simulation in predicting and eliminating defects in sand casting parts.
1. Introduction
The divided axle housing is a critical load-bearing component in tractors, and its quality directly affects the reliability and service life of the vehicle. In the foundry practice, the sand casting of such components is challenging due to complex geometries, variable wall thicknesses, and the inherent shrinkage behavior of steel alloys. According to statistical data, the occurrence rate of shrinkage porosity and shrinkage cavities in the axle housing casting process was as high as 70%, leading to frequent oil leakage problems during service. This resulted in economic losses exceeding one million RMB annually. The primary objective of my research is to solve this persistent quality issue through systematic casting process design and numerical simulation.
With the rapid development of computer simulation technology, casting simulation software such as ProCAST has become an indispensable tool for foundry engineers. By simulating the mold filling and solidification processes, one can predict the locations and sizes of casting defects before actual production, thus enabling process optimization in a virtual environment. This approach significantly reduces trial-and-error costs and improves the quality of sand casting parts. In this study, I used ProCAST to analyze the flow field, temperature field, and defect formation in the axle housing casting, and subsequently optimized the gating and feeding system.
2. Structural and Process Analysis of the Axle Housing
The divided axle housing under investigation has a nominal envelope size of 452 mm × 340 mm × 396 mm. The average wall thickness is 25 mm, with a maximum of 29 mm and a minimum of 15 mm. The material is ZG40Mn, a low-alloy cast steel commonly used for structural components requiring high strength and toughness. Table 1 lists the chemical composition requirements, and Table 2 shows the mechanical properties after heat treatment (normalizing and tempering).
| Element | Content |
|---|---|
| C | 0.35–0.45 |
| Mn | 1.0–1.5 |
| Si | 0.35–0.45 |
| S | ≤0.03 |
| P | ≤0.03 |
| Condition | Yield strength (MPa) | Tensile strength (MPa) | Elongation (%) | Reduction of area (%) | Hardness (HB) |
|---|---|---|---|---|---|
| Normalized + Tempered | ≥295 | ≥640 | ≥12 | ≥30 | 163–200 |
The structural analysis revealed that the axle housing has a relatively symmetrical shape, but with significant variations in wall thickness. The critical functional areas are the flange surfaces and the internal cylindrical cavities, which must be free from defects to ensure proper sealing and fatigue life. During the casting process, I decided to place the component horizontally, as shown in the schematic layout (not reproduced here), with the main cavity oriented downward to ensure soundness in the critical sections. The parting plane was selected at the horizontal symmetry plane, which simplifies mold making and core placement.
The pouring temperature was set to 1580 °C based on a series of simulations that compared 1540 °C, 1560 °C, and 1580 °C. The higher temperature provides better fluidity and promotes a more favorable temperature gradient for directional solidification, while avoiding excessive thermal stresses that could cause hot tearing. The casting shrinkage allowance was determined as 2% according to the recommended values for low-alloy steel in inhibited sand molds. The dimensional tolerance grade was selected as CT12 according to the GB/T 6414 standard. The core design comprised a single complex core with horizontal and vertical core prints. The horizontal core prints were 50 mm long with a taper of 3°, while the vertical upper core print was 50 mm long with a taper of 4°, and the lower core print was 30 mm long with a taper of 1°.
3. Gating System Design
The gating system is the pathway through which molten metal enters the mold cavity. A well-designed gating system must ensure that the metal fills the cavity smoothly, without turbulence, entrainment of gases, or erosion of the mold, while providing a favorable temperature distribution for feeding. Based on the existing process that used a bottom sprue gate and suffered from severe shrinkage defects, I opted for a middle-rail (mid-height) ingate configuration, which combines the advantages of top and bottom gating.
The minimum static pressure head was calculated using the following formula:
$$ h_M = L \tan\alpha \tag{1} $$
where \( h_M \) is the minimum pressure head, \( L \) is the flow distance from the sprue center to the highest point of the casting (320 mm), and \( \alpha \) is the pressure angle (10°). This gives \( h_M = 320 \times \tan 10^\circ \approx 56 \) mm. Adding the height of the casting above the parting line (190 mm), the total required sprue height was determined as 246 mm.
To design the cross-sectional areas of the gating system components, I used the pouring ratio velocity method. The casting mass was calculated as 117 kg (ZG40Mn density 7.8 g/cm³). The choke area (inner gate area) was obtained from the formula:
$$ A_{\text{阻}} = \frac{G_L}{k \cdot t \cdot S’} \tag{2} $$
where \( G_L \) is the metal weight (108 kg for the casting itself), \( k \) is the pouring ratio velocity (0.6 kg/(cm²·s)), \( t \) is the pouring time, and \( S’ \) is the flow coefficient (0.8 for high-manganese steel). The pouring time was found using the empirical relation \( t = C \cdot \sqrt{G_L} \), where \( C = 0.8 \), giving \( t = 0.8 \times \sqrt{108} \approx 8.3 \) s, rounded to 10 s.
Thus the inner gate area became:
$$ A_{\text{ingate}} = \frac{108}{0.6 \times 10 \times 0.8} = 22.5 \ \text{cm}^2 \tag{3} $$
I then designed three gating systems with different area ratios, as summarized in Table 3.
| Scheme | Type | Ingate area (cm²) | Runner area (cm²) | Sprue area (cm²) | Area ratio |
|---|---|---|---|---|---|
| Scheme 1 | Closed | 22.5 | 29.25 | 33.75 | 1 : 1.3 : 1.5 |
| Scheme 2 | Open | 22.5 | 21 | 18.8 | 1.2 : 1.1 : 1 |
| Scheme 3 | Semi-closed | 22.5 | 18 | 27 | 1 : 0.8 : 1.2 |
For Scheme 1 (closed), the sprue diameter was 66 mm, and the runner had a trapezoidal cross-section. For Scheme 2 (open), the sprue diameter was 49 mm. For Scheme 3 (semi-closed), the sprue diameter was 59 mm. All systems used a single sprue, a single runner, and a single ingate. A ceramic foam filter was placed at the bottom of the sprue to trap impurities and stabilize the metal flow.
4. Numerical Simulation of Mold Filling and Solidification
I used the ProCAST finite element software to simulate the filling and solidification processes of the three gating systems. The 3D models were created in Creo and imported into ProCAST in .stp format. The mesh was generated with a maximum element size of 6 mm for the casting and 15 mm for the runner and sprue. The mold was assumed to be a virtual sand box. The material thermal properties were obtained from ProCAST database for ZG40Mn, including thermal conductivity, specific heat, and solid fraction curves. The heat transfer coefficient between casting and sand mold was set to 500 W/(m²·K). The initial temperature of sand was 30 °C, and the pouring temperature was 1580 °C with a pouring time of 10 s.
4.1 Filling Process Analysis
The filling process was simulated for each scheme. I analyzed the velocity field and the solidification fraction during filling. Table 4 summarizes the solidification fraction values at different filling percentages.
| Filling stage | Scheme 1 (closed) | Scheme 2 (open) | Scheme 3 (semi-closed) |
|---|---|---|---|
| 25% | 0% | 0% | 0% |
| 30% | 0% | 0.1% | 0% |
| 64% | 2.5% | 10.5% | 4% |
| 96% | 10.4% | 18.8% | 12.7% |
From the filling simulations, I observed that Scheme 2 (open) had significant metal stream breakage and isolated liquid droplets, leading to a high probability of cold shuts and gas entrapment. The open system also had poor slag retention because the runner was not fully filled. Scheme 1 (closed) had a smoother filling but with a higher velocity and some initial splashing. Scheme 3 (semi-closed) exhibited the most uniform and gentle filling, with no stream breakage, and the runner remained filled throughout the process, providing good slag retention. The semi-closed system thus offered the best filling performance for these sand casting parts.
4.2 Solidification and Temperature Field Analysis
After filling, the solidification process was analyzed up to the complete cooling of the casting. I tracked the evolution of the solid fraction over time. Table 5 shows the solid fraction values at representative times for each scheme.
| Time (s) | Scheme 1 | Scheme 2 | Scheme 3 |
|---|---|---|---|
| ~10 | 16% | 30% | 24% |
| ~30 | 26% | 50.7% | 69.9% |
| ~100 | 74.4% | 78.9% | 77.3% |
| ~120 | 86.6% | 92.6% | 82.7% |
The temperature field results revealed that isolated liquid regions formed in the later stages of solidification for all three schemes. These isolated pools of liquid metal cannot be fed by outer risers, and thus they are prone to shrinkage porosity and cavities. The open system (Scheme 2) produced the largest and most widely distributed isolated liquid regions, which would inevitably lead to severe internal defects. The closed system (Scheme 1) had a moderate amount of isolated liquid, while the semi-closed system (Scheme 3) showed the smallest isolated liquid regions, but still not completely eliminated.
4.3 Shrinkage Defect Prediction
I used ProCAST’s porosity criterion to compute the shrinkage porosity and cavity locations for each scheme. The simulation predicted the locations of defects primarily in the inner wall of the cylinder cavity and at the bottom of the internal oil seal seat area, which matched the actual leak locations observed in production. The defect area percentages are compared in Table 6.
| Scheme | Defect distribution | Severity |
|---|---|---|
| Scheme 1 (closed) | Isolated shrinkage in the internal cavity | Moderate |
| Scheme 2 (open) | Large shrinkage network across the thick sections | Severe |
| Scheme 3 (semi-closed) | Small shrinkage near hot spots | Mild |
Based on these results, I discarded Scheme 2 and proceeded to improve Schemes 1 and 3 by adding risers and chills.
5. Riser and Chill Design for Optimized Feeding
The feeding system is essential to compensate for the volumetric shrinkage of the steel during solidification. My goal was to create a directional solidification gradient from the remotest parts of the casting toward the risers. I computed the modulus of the casting using the formula:
$$ M = \frac{V}{A} \tag{4} $$
where \( V \) is the volume and \( A \) is the cooling surface area. For the maximum hot spot with a wall thickness of 22 mm, the modulus was calculated as \( M = \frac{22 \times 160}{2(22+160)} = 9.7 \) mm. The riser modulus should be larger than the casting modulus. I designed the riser dimensions using:
$$ D_r = (1.2 \text{ to } 2) \times \delta_c \tag{5} $$
$$ H_r = (1.0 \text{ to } 1.5) \times D_r \tag{6} $$
where \( \delta_c \) is the hot spot thickness. Taking \( \delta_c = 22 \) mm, I obtained a theoretical riser diameter of 44 mm and height of 66 mm. However, considering safety and practical limitations, I selected a riser diameter of 80 mm and height of 150 mm, giving a modulus of 18.75 mm, which is well above the required 9.7 mm. I placed one blind riser at the ingate and three open risers on the upper surface of the casting. In addition, chills were installed in the internal cavity and on the thick side walls. The chills were steel blocks of size 30 mm × 50 mm × 60 mm. They enhance local cooling and promote a favorable solidification sequence. Figure 1 (inserted earlier) shows a typical sand casting manufacturer’s facility for such components.
After adding the risers and chills, I re-simulated the improved Schemes 1 and 3. The simulation results for the optimized Scheme 1 still showed some remaining shrinkage, especially near the flange area, whereas the optimized Scheme 3 exhibited no internal shrinkage defects; the porosity was completely eliminated and the defects were transferred to the risers. This indicates that the semi-closed gating system combined with appropriate riser and chill placement produces sound sand casting parts.
6. Determination of Optimal Parameters
Based on the simulation results, I selected the optimized semi-closed gating system as the final casting process. The key parameters are listed in Table 7.
| Parameter | Value |
|---|---|
| Casting material | ZG40Mn |
| Pouring temperature | 1580 °C |
| Pouring time | 10 s |
| Mold method | Water glass sand |
| Number of castings per mold | 1 |
| Gating type | Semi-closed |
| Area ratio (ingate:runner:sprue) | 1 : 0.8 : 1.2 |
| Ingate area | 22.5 cm² |
| Runner area | 18 cm² |
| Sprue area | 27 cm² |
| Riser type | Blind riser at ingate, open risers on top |
| Chill location | Internal cavity and thick wall sections |
This configuration ensures smooth filling, efficient slag trapping, and a proper temperature gradient for directional solidification. The semi-closed gating system combines the advantages of both closed and open systems, making it ideal for medium-sized steel castings such as this axle housing.
7. Experimental Validation
To verify the reliability of the optimized process, I conducted a series of trials in the foundry. I produced several batches of the divided axle housing using the final semi-closed gating system. The castings were then subjected to sectioning inspection and air tightness (air pressure) testing.
7.1 Sectioning Test
I cut the first two castings into 100 mm slices, as shown in the process flow (not illustrated). I examined all critical sections under a 10× magnifying glass. No visible cracks or voids larger than 3 mm were found in the previously defective areas. After five production batches, a total of 630 castings were produced, with a sampling rate of two per hundred for sectioning. I inspected 31 cut sections from these castings, and none of them showed any shrinkage porosity or cavities. This confirms that the optimized process is capable of producing sound sand casting parts.
7.2 Air Tightness Test
Air tightness testing is the most important quality check for the axle housing because any internal porosity can lead to oil leakage in service. I used the direct pressure method: the machined housing was sealed with special fixtures and immersed in water, then pressurized with air at 0.5 MPa for at least 10 minutes. Any bubble formation indicated a defect. I tested 20 castings from two batches (8 from one batch and 12 from another). All 20 samples passed without any leakage. Table 8 shows the test results.
| Sample ID | Test pressure (MPa) | Sealing condition | Dwell time (min) | Leakage (Y/N) | Result |
|---|---|---|---|---|---|
| A1–A8 | 0.5 | Ok | 10 | N | Pass |
| B1–B12 | 0.5 | Ok | 10 | N | Pass |
Up to the end of 2019, more than 6,000 axle housings had been produced using the optimized process. Only 174 failed the quality inspection, which corresponds to a defect rate of approximately 2.9%. This is a dramatic improvement from the original 70% failure rate. The new process not only reduced material waste but also greatly improved the reliability of the final product.
8. Conclusions
Through my research and experiments, I reached the following conclusions:
(1) The filling simulation of the three gating systems demonstrated that the semi-closed gating system (Scheme 3) offered the most stable and uniform metal flow with no turbulence or stream breakage. The runner remained completely filled throughout the process, which improved slag retention and reduced the formation of inclusions.
(2) The solidification simulation revealed that isolated liquid regions inevitably form in the final stages of solidification. For this axle housing, the open gating system (Scheme 2) produced the worst isolated regions and hence the highest shrinkage defect risk. The semi-closed system was the best initial design, but it still required additional feeding measures.
(3) By adding a blind riser at the ingate, three open risers on the top surface, and steel chills in the internal cavity and thick sections, I successfully eliminated the internal shrinkage defects in the optimized semi-closed casting system. The numerical simulation confirmed that all porosity was transferred to the risers, and the casting became fully sound.
(4) Industrial trials verified the simulation results. Sectioning tests showed no internal defects, and air tightness tests on 20 samples all passed at 0.5 MPa. Over 6,000 sand casting parts were produced with a failure rate below 3%, proving the effectiveness and robustness of the optimized process.
This work demonstrates the great value of numerical simulation in the design and optimization of sand casting processes. The methodology can be extended to other complex steel castings to achieve high-quality, defect-free sand casting parts.