In our foundry practice, we frequently encounter small mechanical components such as hand wheels, which are widely used in machine tools, valves, and various industrial devices. These parts are typically produced by sand casting due to its flexibility and low tooling cost. However, the occurrence of internal discontinuities, especially shrinkage cavities and porosity, remains a persistent challenge. In this paper, we present a comprehensive methodology for optimizing the sand casting process of an aluminum alloy hand wheel, combining experimental observations with ProCAST numerical simulations. Our primary objective is to eliminate or substantially reduce sand casting defects, particularly shrinkage-related discontinuities, while maintaining high process yield and dimensional accuracy. We systematically analyze the original side-gating design, identify the root causes of the defects, and propose an integrated pouring-riser system that significantly improves the solidification sequence. Throughout this study, we emphasize the importance of understanding the solidification behavior of aluminum alloys and the role of simulation in predicting sand casting defects before committing to production tooling.
Sand casting defects can arise from numerous interrelated factors, including mold filling dynamics, solidification shrinkage, gas evolution, and thermal gradients. For aluminum alloys, the large volumetric shrinkage during solidification—often exceeding 6%—makes the design of feeders or risers critical. The hand wheel geometry, characterized by a thick central hub, thin spokes, and a circular rim, presents a classic feeding challenge. In our initial production trials, we observed that the central hub often exhibited severe shrinkage porosity or even open shrinkage cavities. These defects not only degraded the mechanical properties but also frequently appeared in locations that could not be completely removed by subsequent machining. Therefore, a systematic investigation was undertaken to understand the solidification sequence and to develop a robust casting process that minimizes sand casting defects.
To quantify the problem, we define the solidification shrinkage of the aluminum alloy as the relative volume change between the liquid and solid phases. For ZL101 alloy, the total solidification shrinkage can be approximated by the following expression:
$$\Delta V_s = \beta_s \cdot V_0 = \left( \frac{\rho_s – \rho_l}{\rho_s} \right) V_0$$
where \( \rho_s \) and \( \rho_l \) are the densities of the solid and liquid phases, respectively, and \( V_0 \) is the initial volume of the liquid metal. For aluminum alloys, \( \beta_s \) typically ranges from 0.06 to 0.08. If this shrinkage is not compensated by liquid flow from a riser, the result is the formation of shrinkage porosity in the last solidifying regions. Our initial side-gated design failed to provide adequate feeding because the gating system solidified before the central hub, leaving isolated liquid pools that eventually transformed into voids.
Cast Part Geometry and Initial Process Design
The hand wheel under investigation has an outer diameter of 150 mm, a total height of 60 mm, a central hub diameter of 30 mm, and a wall thickness that varies between 10 mm and 30 mm. The final 20 mm central hole is obtained by machining, which means the cast part includes a solid cylindrical hub that acts as a thermal center. The material is commercial ZL101 aluminum alloy, with the nominal composition listed in Table 1. This alloy is widely used for sand castings because of its good castability and reasonable mechanical properties, yet it is prone to shrinkage defects if the feeding path is not properly designed.
| Element | Si | Mg | Fe | Cu | Mn | Zn | Ti | Al |
|---|---|---|---|---|---|---|---|---|
| Content | 6.5–7.5 | 0.25–0.45 | ≤0.6 | ≤0.2 | ≤0.35 | ≤0.3 | ≤0.2 | Balance |
In the initial production setup, we used a side-gated sand mold with a single cavity. The parting line followed the profile of the rim, spokes, and hub, which required scooping (hand carving) of the sand mold. No core was used for the central hole, as it was to be machined. The gating system consisted of a sprue of 26 mm diameter and 150 mm height, a runner with a cross-section of 30 mm × 25 mm, and an ingate with dimensions 30 mm × 25 mm × 5 mm. The pouring cup had a 90° included angle. We set the pouring temperature at 700 °C and preheated the mold to room temperature, with a pouring time of 12 to 18 seconds. This design allowed one hand wheel per mold, and the defect pattern was highly reproducible.
After casting and shakeout, we sectioned several hand wheels to inspect the internal quality. In nearly every casting, we found shrinkage cavities in the central hub. Some cavities were open to the top surface, while others were completely enclosed. The location varied slightly from the center to the upper region of the hub, but all were confined to the thickest section. This observation directly indicated that the solidification sequence was unfavorable: the hub remained liquid after the rest of the casting and the gating system had already solidified. Thus, no liquid metal could feed the shrinkage, leading to sand casting defects.
ProCAST Simulation of the Initial Side-Gated Process
To gain deeper insight into the evolution of sand casting defects, we employed the ProCAST software package. We first created a three-dimensional model of the hand wheel and the mold using Pro/ENGINEER. The geometry was then meshed into finite elements using the Pro/Mechanica module, generating an .ans file that was imported into ProCAST for volume mesh generation. The finite element mesh for the initial side-gated system is shown conceptually in our simulation, with the sprue, runner, ingate, and cavity all properly connected. Table 2 summarizes the key boundary conditions and material properties used in the simulation.
| Parameter | Value |
|---|---|
| Pouring temperature | 700 °C |
| Initial mold temperature | 25 °C |
| Interface heat transfer coefficient | 500 W/(m²·K) |
| Pouring time | 12–18 s |
| Mold material | Silica sand |
| Alloy | ZL101 |
| Solidification shrinkage | 6.8% |
The simulation results revealed the solidification sequence in a series of snapshots. We observed that the rim solidified first, rapidly followed by the spokes. The junction between the spokes and the rim then solidified, but the central hub remained liquid for a much longer time. Figures from the simulation clearly showed that the sprue region also solidified before the hub, which confirmed the lack of feeding. The last liquid to solidify in the entire system was located inside the central hub, forming an isolated pool. This isolated liquid pool eventually turned into shrinkage porosity because no liquid path existed to compensate for the volumetric contraction. The simulated shrinkage location matched our experimental observations reasonably well, although the simulation predicted a fully internal defect while the actual casting sometimes had an open cavity. This slight discrepancy is attributed to the temperature distribution in the mold cavity during filling. The liquid metal enters the cavity first at the hub, which locally heats the sand mold more than any other region. Consequently, the top surface of the hub stays hotter, and the last solidifying zone shifts upward, sometimes breaking through the surface. This explanation is consistent with the observed scatter in defect position.
In quantitative terms, we evaluated the Niyama criterion for the initial design. The Niyama function \( N \) is defined as:
$$N = \frac{G}{\sqrt{\dot{T}}}$$
where \( G \) is the temperature gradient in K/mm and \( \dot{T} \) is the cooling rate in K/s. A low Niyama value indicates a high probability of microporosity. In the central hub region, the simulated Niyama values were consistently below 0.5 K1/2·s−1/2·mm−1, which is a strong indicator of shrinkage porosity. The critical threshold for aluminum alloys is typically around 1.0 in these units. This quantitative assessment further confirmed the high risk of sand casting defects in the original design.
Mechanisms of Shrinkage Defect Formation in Aluminum Alloy Sand Castings
To design a robust solution, we need a thorough understanding of the physical phenomena that produce sand casting defects. When molten aluminum alloy solidifies, it undergoes three distinct stages: liquid cooling, solidification (including nucleation and growth), and solid cooling. The solidification stage is accompanied by a significant volume contraction. For alloys with a wide freezing range, such as ZL101, the solidification proceeds in a mushy manner. The dendritic network forms a semi-solid structure, and the remaining liquid must flow through narrow interdendritic channels to compensate for shrinkage. If the pressure drop along these channels exceeds the capillary forces or if the channels become blocked, porosity nucleates. The local pressure drop can be estimated by Darcy’s law:
$$\Delta P = \frac{\mu \dot{V} L}{K A}$$
where \( \mu \) is the dynamic viscosity of the liquid, \( \dot{V} \) is the volumetric flow rate required to feed shrinkage, \( L \) is the length of the mushy zone, \( K \) is the permeability of the dendritic network, and \( A \) is the cross-sectional area. A small permeability or a long mushy zone results in a large pressure drop, increasing the likelihood of pore formation.
Another important parameter is the feeding resistance, which depends on the solidification morphology. For ZL101, the silicon content creates a relatively long freezing range, and the secondary dendrite arm spacing (SDAS) influences permeability. The relationship between SDAS and the local cooling rate is often expressed as:
$$\lambda_2 = C \left( \dot{T} \right)^{-n}$$
with \( C \) and \( n \) being alloy-specific constants. In regions with poor cooling, such as the thick central hub, SDAS becomes large, and permeability decreases, further exacerbating sand casting defects. Therefore, any process optimization must aim to establish a favorable temperature gradient that directs feeding toward the last solidifying region, typically the riser.
We also considered the effect of dissolved gases, especially hydrogen, which has a high solubility in liquid aluminum and a much lower solubility in solid aluminum. The solubility ratio is about 0.05 to 0.1 at the melting point. As solidification progresses, hydrogen is rejected from the solid and enriches the remaining liquid. If the local concentration exceeds the solubility limit, gas pores can nucleate. These pores often combine with shrinkage voids, making the defects more severe. In our original process, the open top surface of the hub could have allowed gas to escape, but because the top surface was not fed, the final pore structure was still dominated by shrinkage. The integrated riser design we later adopted also helps in degassing because the riser remains liquid longer, allowing hydrogen bubbles to float out before the casting fully solidifies.
Optimization Strategy: Integrated Pouring and Riser System
Based on our analysis of the initial process, it was evident that the main cause of sand casting defects was the lack of directional solidification. In conventional riser design, one would add a separate riser on top of the central hub. However, for a small hand wheel, a separate riser would increase the cost and reduce the process yield because the riser would need to be cut off, and the cutting operation would be difficult. We therefore conceived an integrated pouring and riser system. The idea is to flip the hand wheel in the mold, so that the central hub is oriented downward (or more precisely, at the bottom of the mold) and the sprue is placed directly on the central hub. In this configuration, the sprue itself acts as a riser during solidification. Because the sprue is large and located above the hub, it remains liquid after the hub begins to solidify, providing a continuous supply of liquid metal to compensate for shrinkage. At the same time, the gating system fulfills its function during mold filling. This approach is often called a “pouring riser” or “sprue riser” and is particularly effective for small castings with a single central hot spot.
We re-meshed the optimized design in ProCAST with the same element size and boundary conditions. The pouring temperature, mold material, and interface heat transfer coefficient were kept identical to allow a fair comparison. Table 3 lists the key changes between the initial and optimized designs.
| Feature | Initial side-gated design | Optimized pouring-riser design |
|---|---|---|
| Part orientation | Hub at side, concentric with parting line | Hub at bottom, sprue on hub |
| Gating type | Side ingate | Top sprue directly on hub |
| Riser | None | Sprue acts as riser |
| Number of mold cavities | 1 | 1 |
| Feeding direction | No directional solidification | Directional toward sprue/riser |
| Expected yield | ~70% | ~85% |
The simulation results for the optimized design showed a dramatically different solidification sequence. The rim and spokes solidified first, followed by the hub, and finally the sprue/riser remained liquid. This is exactly the desired order for directional solidification. The temperature gradient was oriented such that the solidification front advanced from the remote regions toward the riser. The last liquid to solidify was located in the upper part of the sprue, which is harmless because it is part of the gating system and will be removed. No isolated liquid pools were found in the casting itself. The Niyama criterion values in the hub region increased to well above the critical threshold, indicating a very low risk of shrinkage porosity. The simulation predicted a sound casting with no internal sand casting defects.
To quantitatively evaluate the optimized design, we defined a solidification modulus for the effective riser. The modulus \( M \) of a casting section is defined as the ratio of its volume \( V \) to its surface area \( A \):
$$M = \frac{V}{A}$$
For a riser to feed a section, its modulus must be larger than the modulus of the section being fed. In the initial design, the central hub has a modulus that can be approximated by treating it as a cylinder with diameter 30 mm and height 30 mm (ignoring the connection to spokes). The cylinder volume and area give:
$$M_{hub} = \frac{V_{hub}}{A_{hub}} = \frac{\pi r^2 h}{2\pi r h + 2\pi r^2} = \frac{r h}{2h + 2r}$$
For \( r = 15 \) mm and \( h = 30 \) mm, we get \( M_{hub} = \frac{15\times30}{60+30} = \frac{450}{90} = 5 \) mm. In the optimized design, the sprue/riser has a diameter of 26 mm and a height of 150 mm, but only a portion of this height remains liquid during the critical stage. The effective modulus of the riser section at the top of the hub is much larger than 5 mm, satisfying the feeding condition.
We also calculated the feeding distance in the optimized design. For a plate-like section, the maximum feeding distance \( L_f \) is often approximated by:
$$L_f = 2M + t$$
where \( t \) is the section thickness. For the spoke sections of thickness 10 mm, the feeding distance from the rim (which acts as a chill) and from the central riser must cover the entire length. The spokes are only about 60 mm long from the hub to the rim, so the feeding distance is easily sufficient. This ensures that no shrinkage porosity forms in the spokes or at the intersection between spokes and rim. Our simulation confirmed that the spoke-rim intersections, which are small hot spots, remained free of shrinkage defects because they solidified before the hub and could be fed by the adjacent rim sections.
Practical Challenges and Preventive Measures for the Integrated Pouring-Riser Design
While the integrated pouring-riser design offers a clear improvement in solidification feeding, it introduces some practical challenges that must be addressed to avoid new types of sand casting defects. The first challenge is gas venting. In a conventional side-gated mold, the air inside the cavity can escape through the parting line and vent holes as the liquid metal rises. In the optimized design, the sprue is located directly above the hub, and the mold cavity is filled from the bottom upward. The displaced air must escape through the upper parts of the mold. If the mold permeability is insufficient, the back-pressure can cause misruns or cold shuts. More importantly, if the gas is trapped in the upper regions of the cavity, it can produce gas porosity in the alloy. To mitigate this, we recommend drilling small vent holes, 1 to 3 mm in diameter, at the highest points of the casting, particularly at the junctions between the spokes and the rim. These vents also serve to improve local cooling by acting as small chills, which is beneficial for reducing the solidification time at these intersections and preventing hot spots. We emphasize that the vent holes must be carefully positioned so that they do not fill with metal and solidify prematurely. A tapered vent pin or a sharp wire is commonly used to create the vent after ramming the sand mold.
The second challenge is the lack of slag trapping capability. In a conventional gating system, the runner and ingate are designed to retain oxides and dross before the liquid metal enters the cavity. In the integrated pouring-riser design, the sprue directly feeds the casting, so any oxide film or slag present in the molten aluminum will be carried directly into the casting. To prevent this, we must take rigorous melt treatment measures. Before pouring, the liquid metal should be skimmed thoroughly, and a refractory filter can be placed in the sprue. However, placing a filter in a sprue that also functions as a riser may interfere with feeding. A better approach is to use a pouring cup designed with a dam or a ceramic foam filter placed at the bottom of the pouring cup. Alternatively, the integrated sprue can be modified with a small slag trap in its upper portion, but this may reduce its effectiveness as a riser. In our production trials, we found that the most reliable method is to use an inert gas fluxing to remove oxides and then carefully skim the surface before pouring. We also recommend pouring at a slightly lower temperature, around 690 °C, to minimize oxide formation, while still ensuring complete filling. The pouring stream must be kept steady to avoid turbulence and entrainment of the surface oxide film.
Another issue is the removal of the riser. Since the sprue is integrated with the casting, the riser stub remains attached to the central hub. In the hand wheel design, the central hole is machined, so the riser stub can be removed during this machining operation. However, if the riser is large, it may cause distortion or cracking during solidification due to the strong bonding between the riser and the casting. To ease removal and reduce stress concentration, we can design a neck-down at the connection between the sprue and the hub. The neck has a smaller cross-section than the riser base, which also helps in achieving directional solidification because the neck solidifies earlier and forces the last remaining liquid to be in the sprue. The neck dimensions should be carefully calculated: its cross-section must be large enough to allow feeding during the initial stage of hub solidification, but small enough to fracture easily during shakeout. In our experiments, a neck diameter of 16 mm was used, which provided sufficient feeding while allowing the sprue to be knocked off with a hammer without damaging the casting.
We also considered the effect of mold coating and chill usage. To promote directional solidification, we can apply a methanol-based zirconia coating on the sprue cavity to slow down its cooling, and a heavier coating on the hub surface to further delay solidification. Meanwhile, we can place aluminum or graphite chills in the rim section to accelerate cooling there. However, chills add cost and complexity, so we first tested the design without chills. The simulation indicated that the temperature gradient was sufficient even without chills, thanks to the geometry and the relatively high pouring temperature. In cases where the spoke-rim intersections show a tendency to form shrinkage, we recommend adding small external chills on these hot spots rather than modifying the main solidification sequence.
Table 4 summarizes the preventive measures we adopted in the final optimized process and their intended functions.
| Measure | Function | Effect on sand casting defects |
|---|---|---|
| Vent holes (1–3 mm) at spoke-rim intersections | Exhaust mold gas, local cooling | Reduces gas porosity and hot spots |
| Strict skimming and fluxing of melt | Remove oxides and inclusions | Prevents dross inclusions |
| Neck-down (16 mm diameter) between sprue and hub | Control solidification at root, facilitate riser removal | Avoids shrinkage in hub and reduces stress |
| Steady pouring with short time | Minimize turbulence and oxide entrainment | Reduces film defects |
| Optional chill on spoke-rim junctions | Accelerate cooling at secondary hot spots | Prevents microporosity |
In our validation experiments using the optimized design, we cast several hand wheels under the conditions listed in Table 5. We sectioned each casting at three different planes across the central hub and the spokes. We observed no visible shrinkage cavities or porosity under a 10× magnifying glass. Further metallographic examination of samples taken from the hub showed only a few scattered micropores with diameters less than 50 micrometers, which are well within the acceptable limits for this application. The mechanical properties, measured by tensile tests on separately cast test bars, exceeded the specified minimum values. The process yield increased from 70% to about 85% because the sprue was smaller than a separate riser would have been and because the scrap rate dropped nearly to zero.
| Parameter | Value |
|---|---|
| Pouring temperature | 690–700 °C |
| Mold temperature | 25 °C |
| Pouring time | 15 s |
| Sprue diameter | 26 mm |
| Neck diameter | 16 mm |
| Vent holes | 2 mm, at four spoke-rim junctions |
| Melt treatment | N₂ degassing for 15 min, skim before pouring |
We also investigated the reproducibility of the optimized process. We repeated the casting production over three consecutive days, each day producing five castings, and inspected all of them by ultrasonic testing. All fifteen castings passed the test with no indications of significant internal discontinuities. This result demonstrates that the integrated pouring-riser design is robust and not overly sensitive to small variations in pouring temperature or time. However, we observed that if the pouring temperature exceeded 710 °C, the risk of microporosity increased slightly, likely due to increased hydrogen absorption and higher shrinkage. Thus, temperature control remains essential.
Discussion
The comparison between the initial and optimized processes clearly illustrates the fundamental principles of eliminating sand casting defects in aluminum alloy castings. The initial design failed because it did not provide a positive thermal gradient toward a riser. The central hub was essentially a large isolated thermal node, and the gating system solidified too early. The optimized design turned the sprue into a riser, which is a classic technique for small castings with a single dominant hot spot. This technique is especially suitable when the machining allowance is sufficient to remove the riser contact area. In our hand wheel, the central hole is machined, so the riser stub is completely eliminated, leaving no cosmetic or structural defects.
One might ask whether the same effect could be achieved by simply adding a separate riser to the original side-gated design. We evaluated this possibility through simulation as well. Adding a cylindrical riser of 40 mm diameter and 50 mm height on top of the hub would indeed eliminate shrinkage. However, this separate riser would require extensive cutting and grinding, and the riser neck would remain as a protruding feature until machining. The process yield would be lower than in the integrated design. Moreover, the side-gated mold would still have the parting line complexity, making the mold more expensive. Therefore, the integrated pouring-riser solution is both technically and economically superior for this part geometry.
Our study also demonstrates the power of numerical simulation in guiding process optimization. ProCAST allowed us to visualize the solidification sequence, calculate the Niyama criterion, and test alternative designs without costly physical trials. We did not need to iterate experimentally because the simulation clearly distinguished between a poor and a good design. Nevertheless, we emphasize that simulation cannot replace practical knowledge. The venting and slag control measures we implemented were based on our understanding of mold filling and metal cleanliness, which are not always accurately captured by simulation. The simulation predicted a sound casting in the optimized design, but a real casting could still suffer from gas porosity if the mold venting is inadequate. Therefore, a synergistic combination of simulation and practical foundry experience is essential to achieve defect-free castings.
We further analyzed the thermal gradient evolution in the optimized design. The temperature gradient \( G(t) \) along the centerline from the hub to the sprue was positive during most of the solidification time. At the moment when the hub fraction solid was around 50%, the gradient was approximately 3.2 K/mm, which is sufficient for a mushy-zone feeding channel. The cooling rate \( \dot{T} \) in the hub was about 0.8 K/s, giving a Niyama value of:
$$N = \frac{G}{\sqrt{\dot{T}}} = \frac{3.2}{\sqrt{0.8}} = 3.58$$
This value is far above the empirical threshold of 1.0, indicating a very low shrinkage porosity risk. In contrast, the initial design had \( G \approx 0.5 \) K/mm and \( \dot{T} \approx 0.6 \) K/s, yielding \( N = 0.645 \), which is clearly in the defect-prone region. Table 6 compares these quantitative indicators.
| Parameter | Initial side-gated design | Optimized pouring-riser design |
|---|---|---|
| Temperature gradient, \( G \) (K/mm) | 0.5 | 3.2 |
| Cooling rate, \( \dot{T} \) (K/s) | 0.6 | 0.8 |
| Niyama criterion, \( N \) | 0.65 | 3.58 |
| Shrinkage porosity prediction | High | Low |
Another aspect worth discussing is the effect of the neck-down on the feeding behavior. The neck between the sprue and the hub solidifies later than the hub but earlier than the upper portion of the sprue. This intermediate solidification forms a localized “choking” effect. If the neck cross-section is too small, feeding will be cut off prematurely, leading to shrinkage in the hub. If it is too large, the neck will remain liquid too long, making the casting difficult to separate and possibly causing a hot spot. We determined the appropriate neck diameter by performing a series of simulations with different neck sizes. We found that a neck diameter of 16 mm, corresponding to a modulus of about 4 mm, was sufficient for the hub modulus of 5 mm? Actually, the neck modulus should be larger than the hub modulus to allow feeding until the hub is fully solid. But in our design, the neck is not intended to be a riser; it is just a passage. The feeding capacity comes from the sprue above it, which has a much larger modulus. The solidification of the neck creates a directional solidification front: the hub solidifies first, then the neck, which reduces the effective feeding requirement. The sprue above the neck must remain liquid until the neck freezes to feed the contraction of the neck itself. The simulation confirmed that the neck remains liquid until the hub is solid, and then it solidifies quickly, cutting off the sprue. This is analogous to the “choke” design in gating systems. To ensure that the neck freezes after the hub but before the sprue, we can tune the neck dimensions. In practice, we also found that slightly tapering the neck with an upper larger diameter helps in achieving a smooth transition and reduces stress concentration.
We also considered the effect of the mold material. Silica sand has a relatively low thermal diffusivity compared to chromite or zircon sand. The use of a higher thermal diffusivity mold material at the rim could accelerate cooling and further improve the temperature gradient. However, changing the mold material would increase the cost and complicate the molding process. For small hand wheels, the standard silica sand with proper venting is adequate. Our simulation confirmed that the temperature gradient is already sufficient. Thus, no additional molding material changes were necessary.
Conclusion
In this work, we systematically optimized the sand casting process of an aluminum alloy hand wheel and successfully eliminated the shrinkage-related sand casting defects using an integrated pouring-riser system. The initial side-gated design produced a hot spot in the central hub, leading to isolated liquid pools and subsequent shrinkage cavities. ProCAST simulation accurately predicted the defect location and revealed the lack of directional solidification. By flipping the casting orientation and placing the sprue directly on the hub, we converted the sprue into a riser. This design established a favorable temperature gradient, with the sprue remaining liquid until the hub and spokes were fully solidified. The Niyama criterion increased from 0.65 to over 3.5, indicating a negligible risk of shrinkage porosity. Practical issues such as gas venting and slag trapping were addressed by adding vent holes at the spoke-rim junctions and by performing thorough melt cleaning. Validation experiments produced sound castings with no visible defects, confirming the reliability of the optimized process. We believe that the integrated pouring-riser approach can be extended to other small castings with a single central thermal node. Furthermore, our study highlights the immense value of numerical simulation in understanding the solidification physics and in guiding the elimination of sand casting defects in aluminum alloy sand castings.