As a researcher in advanced manufacturing technologies, I have been deeply involved in the development of sand casting parts, particularly those with complex geometries and high-performance requirements. In this study, I focus on the casting process of a double-suction impeller made from C95820 aluminum bronze, leveraging the advantages of sand mold 3D printing. This impeller is a critical component in turbomachinery used for high-flow and high-pressure fluid handling systems, such as pumps and compressors. Compared to traditional single-suction impellers, double-suction designs offer greater flow capacity and higher efficiency, making them essential in industries like energy, petrochemicals, and water treatment. The production of such sand casting parts, however, poses significant challenges due to material properties and structural complexities, which I aim to address through innovative process design.
Aluminum bronze, specifically C95820, is known for its narrow crystallization temperature range, good fluidity, and resistance to segregation and dispersed shrinkage, making it suitable for dense castings. It is commonly used in gears, turbines, bushings, and propellers. However, during melting and pouring, aluminum bronze tends to form oxide inclusions and gas porosity due to its high affinity for oxygen, and it exhibits significant shrinkage during solidification, leading to concentrated shrinkage cavities and slag inclusions. These issues necessitate precise control over the casting process to ensure the integrity of sand casting parts. Traditional sand casting methods often struggle with such demands, especially for intricate designs, prompting the adoption of 3D printing technology.
3D printing has revolutionized sand casting by offering high-quality, short-cycle, and low-material-consumption alternatives to conventional methods. It eliminates the limitations of traditional molding, allowing for free design of sand cores and gating systems. The simplification of sand mold production, reduced lead times, and lower costs make it ideal for complex sand casting parts like the double-suction impeller. In this work, I combine sand mold 3D printing with simulation-driven process optimization to achieve a defect-free casting.
The double-suction impeller under study features a complex structure with asymmetric double-layer blades, as shown in the following image. I inserted this visualization to illustrate the intricate geometry that challenges traditional casting. The impeller has a maximum outer diameter of 1015 mm, a height of 630 mm, and a central through-hole, with uneven wall thicknesses ranging down to a minimum of 5 mm. The inner cavity is particularly complex, with two layers of blades arranged non-symmetrically, connected by a thin plate only 15 mm wide. Additionally, the upper and lower cover plates are linked to the blade layers via 19-mm-wide connection plates. This design increases the risk of sand core damage during demolding and core removal in conventional processes, and the aluminum bronze material exacerbates defects like oxide inclusions and gas pores. Therefore, I opted for a sand mold 3D printing approach to overcome these hurdles.
To begin, I conducted a thorough analysis of the impeller’s structural characteristics. The asymmetry and thin connections necessitate a robust casting process to prevent defects. I formulated the chemical composition of C95820 aluminum bronze, as detailed in Table 1, which serves as a reference for material behavior during casting. This composition influences fluidity, shrinkage, and defect formation, all critical for producing high-quality sand casting parts.
| Element | Content |
|---|---|
| Cu | ≥77.5 |
| Al | 9.0–10.0 |
| Ni | 4.5–5.8 |
| Fe | 4.0–5.0 |
| Mn | ≤1.5 |
In designing the casting process, I prioritized the gating and riser system to ensure smooth filling and effective feeding. Based on principles of casting position, I selected a bottom-gating approach to minimize turbulence and promote directional solidification. This is crucial for aluminum bronze sand casting parts to avoid oxide inclusion and gas entrapment. The gating system consists of four bottom gates directly connected to four blind risers, as depicted in the 3D model. This design facilitates controlled metal flow and enhances riser feeding capability. To optimize the system, I employed fluid dynamics equations to estimate flow velocities and pressure drops. For instance, the velocity of molten metal in the gates can be approximated using the Bernoulli equation for incompressible flow:
$$v = \sqrt{\frac{2(P_1 – P_2)}{\rho} + 2gh}$$
where \(v\) is the velocity, \(P_1\) and \(P_2\) are pressures at different points, \(\rho\) is the density of aluminum bronze (approximately 7600 kg/m³), \(g\) is gravitational acceleration, and \(h\) is the height difference. This helps in designing gates to maintain laminar flow and reduce oxidation.
Riser design was guided by simulation results from ProCAST software. Initial simulations without risers revealed shrinkage porosity in thick sections like blade roots and tails, as well as connection plates. I iterated through multiple designs, ultimately incorporating six top elliptical open risers for the upper cover plate, a central cylindrical open riser for the filled through-hole, small cylindrical risers at each blade tail for feeding and impurity removal, and four symmetric blind risers at the bottom for the lower cover plate. This configuration ensures sequential solidification, with risers solidifying last to compensate for shrinkage. The effectiveness of risers can be evaluated using the modulus method, where the riser modulus \(M_r\) should exceed the casting modulus \(M_c\):
$$M = \frac{V}{A}$$
where \(V\) is volume and \(A\) is surface area. For the impeller’s thick sections, I calculated moduli to size risers appropriately, as summarized in Table 2. This mathematical approach minimizes trial-and-error in producing sand casting parts.
| Section | Volume (cm³) | Surface Area (cm²) | Modulus (cm) | Riser Type |
|---|---|---|---|---|
| Upper Cover Plate | ~15,000 | ~5,000 | 3.0 | Elliptical Open |
| Blade Root | ~500 | ~300 | 1.67 | Small Cylindrical |
| Lower Cover Plate | ~12,000 | ~4,500 | 2.67 | Blind |
| Central Hub | ~8,000 | ~3,000 | 2.67 | Cylindrical Open |
Simulation with ProCAST played a pivotal role in validating the process. I set up parameters reflective of actual conditions: mold material as resin-coated sand, pouring temperature at 1150°C, initial mold temperature at 80°C, heat transfer coefficient between casting and mold at 500 W/m²·K, and pouring time of 120 s. The filling simulation showed metal entering through the gates into blind risers before rising steadily into the cavity, with no splashing or counterflow. Velocity magnitudes remained below critical thresholds, typically under 15 m/s, to prevent turbulence. The filling sequence is described by the continuity equation:
$$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0$$
where \(\rho\) is density and \(\mathbf{v}\) is velocity vector. This ensures mass conservation during filling. The solidification simulation revealed that thin sections like blades solidified first, followed by thicker areas, with risers solidifying last—a pattern confirming directional solidification. The solid fraction evolution over time can be modeled using the heat conduction equation:
$$\rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + L \frac{\partial f_s}{\partial t}$$
where \(T\) is temperature, \(c_p\) is specific heat, \(k\) is thermal conductivity, \(L\) is latent heat, and \(f_s\) is solid fraction. Defect prediction indicated shrinkage porosity concentrated in risers and gating systems, with no defects in the impeller itself, verifying the design’s efficacy for sand casting parts.
Moving to sand mold fabrication, I exploited the design freedom of 3D printing to create integrated sand cores for the critical blade structures. Traditional methods would require multiple core pieces, risking misalignment and accuracy loss, but 3D printing allows monolithic cores that preserve dimensional precision. I designed the sand mold with alignment features and parting lines to ensure easy assembly. Additionally, to address the high gas evolution from resin-coated sand—a common issue in sand casting parts—I incorporated ventilation channels. For simple sections, straight channels sufficed, but for curved blade regions, I designed conformal channels following the contour of the cores. This enhances gas expulsion during pouring, reducing porosity risk. The permeability of these channels can be estimated using Darcy’s law:
$$Q = \frac{kA}{\mu} \frac{\Delta P}{L}$$
where \(Q\) is gas flow rate, \(k\) is permeability, \(A\) is cross-sectional area, \(\mu\) is gas viscosity, \(\Delta P\) is pressure difference, and \(L\) is channel length. Conformal channels offer larger \(A\) and shorter \(L\), improving \(Q\).
The printing process utilized a self-developed 3D printer with parameters optimized for sand casting parts, as listed in Table 3. These settings balance speed, resolution, and strength to produce durable molds capable of withstanding molten metal pressures.
| Parameter | Value |
|---|---|
| Laser Power | 900 W |
| Beam Diameter | 1.1 mm |
| Scanning Speed | 2900 mm/s |
| Layer Thickness | 0.3 mm |
| Material | Resin-Coated Ceramic Sand |
Post-printing, I cured the sand molds by flame-sealing the surface and oven-heating to 180°C, followed by slow cooling. This strengthens the molds and reduces residual moisture. Assembly involved stacking the molds with precise alignment, and I inserted vent ropes into the channels to facilitate gas escape during pouring. For casting, I preheated the mold cavity to 80°C, surrounded it with water-glass sand for support, and poured aluminum bronze at 1150°C over 120 seconds. The vent ropes were ignited to burn off gases, ensuring a smooth pour. After cooling, I removed the casting, cut off the risers and gating system, and inspected the impeller. The result was a sound sand casting part with no visible defects, meeting quality standards.
To further analyze the process, I delved into the thermophysical properties of aluminum bronze that influence sand casting parts. The latent heat of fusion \(L_f\) for C95820 is approximately 220 kJ/kg, and the solidus and liquidus temperatures are around 1040°C and 1080°C, respectively. The narrow freezing range \(\Delta T = T_{liquidus} – T_{solidus}\) of about 40°C contributes to good fluidity but also rapid solidification, necessitating efficient riser feeding. The feeding distance \(F_d\) can be estimated using empirical formulas for sand casting parts:
$$F_d = k \sqrt{t}$$
where \(k\) is a material constant (about 2.5 for aluminum bronze) and \(t\) is section thickness. For the 19-mm connection plates, \(F_d\) is roughly 11 mm, justifying the close riser placement.
In terms of fluid flow during filling, the Reynolds number \(Re\) indicates flow regime:
$$Re = \frac{\rho v D}{\mu}$$
where \(D\) is hydraulic diameter and \(\mu\) is dynamic viscosity (about 0.005 Pa·s for aluminum bronze at 1150°C). With \(v\) kept below 1 m/s in the cavity, \(Re\) remains under 2000, ensuring laminar flow and minimizing oxide formation. This is critical for aluminum bronze sand casting parts due to their oxidation tendency.
The solidification time \(t_s\) for sand casting parts can be approximated using Chvorinov’s rule:
$$t_s = C \left( \frac{V}{A} \right)^n$$
where \(C\) and \(n\) are constants dependent on mold material and casting conditions. For resin sand molds, \(C\) is about 1.5 min/cm² and \(n\) is 2. For the impeller’s thickest section (modulus 3.0 cm), \(t_s\) is approximately 13.5 minutes, aligning with simulation results. This guides riser design to remain liquid longer.
Gas porosity prevention is another key aspect. The amount of gas generated from resin sand can be quantified by the decomposition reaction, releasing gases like CO and CO₂. The volume \(V_g\) of gas produced per unit mass of sand is roughly 0.1 m³/kg at casting temperatures. With conformal channels, the expulsion efficiency \(E\) is enhanced, reducing porosity volume \(V_p\):
$$V_p = V_g (1 – E)$$
By designing channels with high \(E\) (estimated >90%), \(V_p\) becomes negligible, ensuring dense sand casting parts.
In conclusion, this study demonstrates a comprehensive approach to producing complex sand casting parts like the double-suction impeller via sand mold 3D printing. By integrating simulation-driven design of gating and riser systems, I achieved smooth filling and directional solidification, confining defects to expendable sections. The use of 3D printing enabled monolithic sand cores with conformal ventilation, overcoming traditional limitations. The successful casting validates the methodology, highlighting the synergy between additive manufacturing and foundry techniques for high-performance sand casting parts. Future work could explore optimization algorithms for riser placement or material innovations for enhanced properties. Overall, this research contributes to advancing sand casting technology, making it more adaptable to demanding applications in turbomachinery and beyond.
Throughout this process, I emphasized the importance of meticulous design and validation for sand casting parts, especially those made from challenging materials like aluminum bronze. The combination of 3D printing and simulation not only improves quality but also reduces lead times and costs, paving the way for more efficient production of intricate sand casting parts. As I reflect on this project, it is clear that such interdisciplinary approaches are essential for pushing the boundaries of manufacturing, ensuring that sand casting parts meet ever-higher standards of performance and reliability.