Our work began when the mine machinery repair factory needed to produce replacement coupling housings for oil pressure torque converters used in coal mine belt conveyors. These components frequently fail under harsh underground conditions, requiring periodic replacement. The housing is a critical safety part: it must withstand an internal oil pressure of six atmospheres during service, and each unit is subjected to an eight‑atmosphere water pressure test before assembly. Any leakage or rupture is unacceptable. The original design called for aluminum alloy sand casting, but the factory – a typical sand casting foundry with limited resources – had no expensive metal molds. We therefore committed to developing a reliable sand casting process for this complex part.
The housing casting is characterized by numerous dispersed thermal nodes, as shown in the figure. These nodes arise from features that cannot be cast in: an oil fill hole, a relief valve hole, intentionally thickened balancing pads, and the inner corner of the coupling bore. In a conventional sand casting foundry, such isolated hot spots are difficult to feed because they solidify last, and the aluminum alloy’s large liquid shrinkage often leaves micro‑porosity. Early trials used the parting line and gating system shown in the sketch: a central sprue with four radial ingates. The initial gating system was very thin – barely able to guide the melt – and the castings repeatedly leaked under water pressure, especially at the four hot‑spot locations. Sometimes water appeared on the outer surface even before the pressure rose.
We analyzed the root cause. The leakage originated from shrinkage porosity in the thermal nodes. Two main factors contributed: first, these regions solidified last and could not be fed by neighboring sections; second, the sand mold’s slow cooling rate expanded the mushy zone, making it difficult to achieve a dense microstructure. As a sand casting foundry we could not afford to switch to metal molds, so we pursued three fundamental strategies: (1) accelerate solidification at the hot spots using chills; (2) enhance feeding wherever possible; (3) redesign local geometries to reduce the thermal mass of the nodes. Based on these principles, we adopted a set of specific measures.
Gating System Enlargement
We retained the original parting line and pouring position but significantly increased the dimensions of the gating system. The original sprue diameter was 20 mm and the ingate cross‑section was only 15 mm × 4 mm, which could only guide the melt. We enlarged the sprue to 40 mm diameter and widened the ingates to 25 mm × 8 mm. The new system acted as an integrated feeder, supplying liquid metal to the central hot spot (node A) during solidification. Table 1 compares the original and modified dimensions.
| Parameter | Original | Modified |
|---|---|---|
| Sprue top diameter (mm) | 20 | 40 |
| Sprue bottom diameter (mm) | 18 | 36 |
| Ingate width (mm) | 15 | 25 |
| Ingate height (mm) | 4 | 8 |
| Number of ingates | 4 | 4 |
| Total ingate area (mm²) | 240 | 800 |
The enlarged gating system significantly improved feeding efficiency. In early trials with the thin system, even placing a 10‑mm‑thick steel external chill on node A could not prevent leakage: water appeared at just 1.2 atm. After modification, node A remained leak‑tight up to 8 atm without any external chill, proving that the feeder effect was more reliable than an external chill that is directly washed by hot metal. In a sand casting foundry, the temperature of an external chill rises rapidly during pouring, reducing its chilling power.
Geometry Modification of Node C
Node C was a deliberate thickening for dynamic balance adjustment. Its original rectangular cross‑section (50 mm × 20 mm, giving a 1000 mm² area) created a large thermal mass. We reshaped it into a tapered section with a maximum thickness of 14 mm and a width of 40 mm (area 560 mm²), as listed in Table 2. This brought the local cooling rate closer to that of the surrounding wall, eliminating the shrinkage porosity that previously caused leakage in that region.
| Dimension | Original | Modified |
|---|---|---|
| Width (mm) | 50 | 40 |
| Maximum thickness (mm) | 20 | 14 |
| Cross‑sectional area (mm²) | 1000 | 560 |
| Estimated modulus (V/A, mm) | ~7.7 | ~5.3 |
The local modulus reduction changed the solidification sequence. Using Chvorinov’s rule, the solidification time is proportional to the square of the modulus:
$$ t_f = C \left( \frac{V}{A} \right)^2 $$
where \(V/A\) is the modulus, and \(C\) is a mold constant. By decreasing the modulus from ~7.7 mm to ~5.3 mm, the solidification time was reduced by about 53%, bringing node C into line with the rest of the casting. In a sand casting foundry, such geometry adjustments are often the most cost‑effective way to eliminate hot spots.
Internal Chills for Nodes B and D
Nodes B and D are small holes that must be drilled later. In the casting, these locations are solid metal that forms isolated hot spots. We placed internal chills – an aluminum rod for node B and a steel rod for node D – each 2 mm smaller than the final hole diameter. The chills were positioned in the mold cavity before pouring. During solidification, they provided intense local cooling. After machining, the chills were drilled out, leaving the required holes. For the steel chill, we pre‑drilled a 3‑mm central hole to prevent drill wandering. Table 3 summarizes the chill properties.
| Node | Chill material | Chill diameter (mm) | Hole final diameter (mm) | Chill length (mm) |
|---|---|---|---|---|
| B | Aluminum alloy | 16 | 20 | 25 |
| D | Steel | 12 | 16 | 20 |
The effectiveness of internal chills can be estimated by the heat extraction capacity. For a steel chill, the heat required to raise it from ambient to solidus temperature is:
$$ Q_{\text{chill}} = m_{\text{chill}} \cdot c_{\text{steel}} \cdot (T_{\text{solidus}} – T_{\text{ambient}}) $$
where \(m_{\text{chill}}\) is the mass, \(c_{\text{steel}} \approx 460 \, \text{J/(kg·K)}\), and the temperature rise is about 550 °C. This heat is drawn from the surrounding liquid, accelerating local solidification. In our trials, internal chills eliminated leakage at nodes B and D without any external cooling. The sand casting foundry found this solution simple and reliable, though care is needed to ensure the chills do not shift during pouring.
Melt Treatment and Pouring Temperature
Porosity in aluminum alloy castings is aggravated by dissolved hydrogen. We implemented rigorous degassing:
$$ \text{Degassing: } 2\text{Al} + 3\text{H}_2\text{O} \rightarrow \text{Al}_2\text{O}_3 + 3\text{H}_2 \uparrow $$
and used rotary degassing with argon at 0.8 L/min for 10 minutes per 100 kg melt. The hydrogen content was reduced from 0.35 mL/100g to below 0.15 mL/100g. Additionally, we lowered the pouring temperature from the original 750 °C to 710 °C, reducing liquid shrinkage and the tendency for hot‑spot porosity. The lower temperature also improved the thermal gradient. Table 4 shows the melt parameters.
| Parameter | Before improvement | After improvement |
|---|---|---|
| Pouring temperature (°C) | 750 | 710 |
| Hydrogen content (mL/100g Al) | 0.35 | 0.12 |
| Degassing method | None | Argon rotary, 10 min |
| Modification | None | Al–5Ti–1B, 0.2% |
Experimental Results
We produced four batches of castings, each comprising two units (total eight castings). The first batch used the original thin gating system and high pouring temperature; both castings leaked at node A under 1.5 atm. The subsequent three batches (batches 2–4) incorporated all the improvements described above – enlarged gating, reshaped node C, internal chills for nodes B and D, degassing, and controlled pouring temperature. All six castings from these batches passed the 8‑atm water pressure test without any leakage. Table 5 summarizes the test results. The sand casting foundry continued to use the process, and later, with further optimization of the alloy composition, the castings even withstood 10 atm.
| Batch | Number of castings | Improvements applied | Test pressure (atm) | Leakage (Y/N) |
|---|---|---|---|---|
| 1 | 2 | Original gating, no chills, 750 °C | 1.5 | Y (node A) |
| 2 | 2 | Enlarged gating + node C reshaped + internal chills + degas + 710 °C | 8 | N |
| 3 | 2 | Same as batch 2 | 8 | N |
| 4 | 2 | Same as batch 2 | 8 | N |
These results confirm that a carefully designed sand casting process can produce pressure‑tight aluminum alloy components, even with complex dispersed hot spots. The key lessons for any sand casting foundry tackling similar parts are: (1) the feeding system should be oversized to act as a feeder; (2) external chills placed in direct contact with the melt stream are ineffective – internal chills or geometry changes are preferable; (3) reducing local thickness to match the surrounding modulus is a powerful tool; (4) clean, degassed melt and lower pouring temperatures minimize residual porosity.
We also observed that the enlarged gating system did not cause any filling problems at the reduced pouring temperature of 710 °C. The fluidity of the aluminum alloy was still adequate to fill the mold completely. This is because the increased ingate area compensated for the lower superheat. The sand casting foundry can therefore safely adopt such modifications without risking misruns.
Conclusion
This study demonstrates that a standard sand casting foundry, equipped only with conventional sand molds and simple steel chills, can manufacture aluminum alloy oil pressure torque converter coupling housings that meet stringent water pressure test requirements. By systematically addressing the solidification issues through gating redesign, local geometry modification, internal chills, melt treatment, and controlled pouring, we eliminated the leakage problem that had plagued earlier attempts. The process is now used in regular production, providing a cost‑effective alternative to metal‑mold casting for small‑to‑medium batch sizes. Our experience serves as a practical guide for other sand casting foundry operations facing similar challenges with pressure‑tight aluminum castings.