In this paper, I present a comprehensive study on the investment casting process of an AlSiCuSc alloy turbine volute using numerical simulation. The turbine volute is a critical pressure-bearing component of a turbocharger, operating under high temperatures, high pressures, and cyclic thermal loads. The quality of this component directly influences the reliability and lifespan of the entire turbocharger system. Investment casting is a preferred manufacturing route for such complex thin-walled parts due to its ability to produce intricate geometries with high dimensional accuracy and excellent surface finish. However, improper process design can lead to severe casting defects such as shrinkage cavities, micro-porosity, hot tearing, and cracks, especially in regions with thick sections and thermal nodes. The traditional trial-and-error approach is costly and time-consuming, and it fails to predict internal defects accurately. Therefore, I employed the ProCAST software to simulate and optimize three different gating and riser system designs for the AlSiCuSc alloy turbine volute. By analyzing mold filling, solidification sequence, isolated liquid pools, and porosity distributions, I identified the optimal process scheme and validated it through experimental casting trials. The results show excellent agreement between simulation and experiment, confirming the reliability of the numerical approach in guiding the investment casting process for high-quality turbine volutes.
Introduction
Turbocharging technology is a key innovation for increasing internal combustion engine power, improving thermal efficiency, and reducing exhaust emissions. The turbine volute is at the heart of this system, enduring extreme conditions that require materials with superior high-temperature mechanical properties and structural integrity. AlSiCuSc alloy, a novel aluminum-silicon-copper-scandium alloy, offers excellent castability, high strength at elevated temperatures, and good corrosion resistance, making it an ideal candidate for turbine volute applications. Nevertheless, the complex geometry of the volute—featuring uneven wall thickness, a spiral flow channel, and thick flanges—poses significant challenges in investment casting. The alternating thick and thin sections create multiple hot spots where molten metal solidifies last, leading to shrinkage porosity if adequate feeding is not provided. In addition, the intricate shape can cause turbulent flow during filling, entrapping gas and forming oxide inclusions.
Investment casting, also known as lost-wax casting, is a near-net-shape process that can produce parts with complex internal and external features. Despite its advantages, the process relies heavily on proper gating and riser design to ensure directional solidification and defect-free castings. Over the years, numerical simulation has emerged as a powerful tool to visualize the casting process and predict defects before actual production. By using software like ProCAST, engineers can simulate mold filling, temperature evolution, solidification kinetics, and porosity formation using criteria such as the Niyama criterion. This virtual prototyping reduces the need for physical trials, shortens development cycles, and lowers costs. In this study, I applied these techniques to three different investment casting designs for the AlSiCuSc alloy turbine volute. I compared the simulation results regarding filling patterns, isolated liquid regions, and shrinkage porosity, and selected the best design for experimental validation.
The structure of this paper is as follows: I first describe the geometry of the turbine volute and the three proposed casting process schemes. Then I present the numerical simulation setup, including material properties, boundary conditions, and computational parameters. After that, I discuss the simulation results for each scheme, focusing on mold filling behavior, solidification progression, and defect prediction. Finally, I report the experimental verification of the optimal scheme and draw conclusions on the effectiveness of the numerical approach for investment casting optimization.
Geometry and Casting Process Design
Geometry Analysis of the Turbine Volute
The turbine volute is a small-to-medium sized structural part with overall dimensions of 134 mm × 129 mm × 104 mm. Its volume is 298,738 mm³ and the mass is 0.83 kg (for the AlSiCuSc alloy with density approximately 2.78 g/cm³). The part exhibits significant wall thickness variation: the maximum wall thickness is 16 mm at the flange and boss regions, while the average thickness is 5.54 mm. The spiral duct wall thickness is only 3 mm. Such abrupt changes in thickness create multiple thermal nodes that are prone to porosity if not properly fed. The geometry also includes a large flat flange on one end and a small cylindrical flange on the other, connected by a curved volute body. This uneven mass distribution leads to non-uniform cooling and solidification, making it challenging to achieve directional solidification.
The existence of thick sections (e.g., the large flange area) requires additional risers or chill blocks to promote a favorable temperature gradient. When the metal solidifies, the thinner sections cool faster, while the thicker ones remain liquid longer. If the liquid in the thick sections is not connected to a riser with sufficient liquid metal, the solidification shrinkage cannot be compensated, resulting in internal cavities. Therefore, the design of an effective gating and riser system is crucial for sound castings.
Design of Three Casting Process Schemes
Based on the geometric features, I designed three different investment casting layouts. The primary goal was to promote directional solidification from the thin sections toward the risers, ensuring that the last solidifying regions are within the risers rather than in the casting itself. All three schemes used a bottom gating system to avoid splashing and mold erosion, with a pouring temperature of 710 °C and a pouring time of 15 seconds. The shell mold was assumed to be preheated to 300 °C. Table 1 summarizes the key dimensions and process yields of the three designs.
| Scheme | Overall Size (mm) | Process Yield (%) | Number of Risers | Riser Type |
|---|---|---|---|---|
| Scheme 1 | 361 × 179 × 359 | 8.6 | 3 | Top cylindrical risers |
| Scheme 2 | 367 × 179 × 282 | 13.1 | 2 | Side riser + top riser |
| Scheme 3 | 331 × 151 × 359 | 12.6 | 2 | Top risers with chill |
In Scheme 1, three cylindrical risers were placed on the thickest parts of the casting: one on the large flange, one on the small flange, and one on the central boss. The overall height of the mold was 359 mm, and the process yield (casting weight divided by total poured weight) was only 8.6%, indicating a large amount of metal consumed by risers. Scheme 2 used two risers: a side riser attached to the large flange and a top riser on the small flange. The overall mold height was reduced to 282 mm, improving the yield to 13.1%. Scheme 3 also used two top risers but incorporated a metallic chill block on the underside of the large flange to accelerate local cooling, resulting in a yield of 12.6%. The process yield in Scheme 1 was the lowest, but it provided the most conservative feeding. All three designs were simulated using ProCAST to evaluate their effectiveness.
Numerical Simulation Methodology
Material Properties of AlSiCuSc Alloy
The chemical composition of the AlSiCuSc alloy used in this study is listed in Table 2. The alloy contains 7.5% Si, 1.7% Cu, 0.6% Sc, 0.4% Mg, 0.15% Ti, and trace amounts of Mn, Zn, Be, and other impurities. The Sc addition refines the grain structure and improves mechanical properties at elevated temperatures. The thermophysical properties required for simulation—such as thermal conductivity, density, enthalpy, solid fraction, and dynamic viscosity—were calculated using ProCAST’s material database and are shown in the following equations and tables.
| Si | Cu | Sc | Mg | Ti | Mn | Zn | Be | Al |
|---|---|---|---|---|---|---|---|---|
| 7.5 | 1.7 | 0.6 | 0.4 | 0.15 | 0.1 | 0.1 | 0.07 | Balance |
The thermal conductivity $$k(T)$$ (W/m·K) and density $$\rho(T)$$ (kg/m³) vary with temperature and are approximated by polynomial fits. For example, the density in the solid state can be expressed as:
$$
\rho(T) = 2.78 \times 10^3 – 0.21 \, (T – 25) \quad \text{for} \quad 25\,^{\circ}\mathrm{C} \le T < T_{\mathrm{solidus}}
$$
and for the liquid state:
$$
\rho(T) = 2.48 \times 10^3 – 0.35 \, (T – T_{\mathrm{liquidus}}) \quad \text{for} \quad T_{\mathrm{liquidus}} \le T \le 710\,^{\circ}\mathrm{C}
$$
The solid fraction $$f_s(T)$$ during solidification was modeled using the Scheil equation:
$$
f_s = 1 – \left(\frac{T_m – T}{T_m – T_{\mathrm{liquidus}}}\right)^{\frac{1}{k_0 – 1}}
$$
where $$T_m$$ is the melting point of pure aluminum (660 °C), $$T_{\mathrm{liquidus}}$$ is the liquidus temperature of the alloy (approx. 610 °C), and $$k_0$$ is the partition coefficient (taken as 0.17 for Al-Si system). The enthalpy $$H(T)$$ (kJ/kg) was calculated by integrating the specific heat capacity and latent heat:
$$
H(T) = \int_{25}^{T} c_p(\tau)\, d\tau + L \cdot f_L(T)
$$
where $$L$$ is the latent heat of fusion (389 kJ/kg) and $$f_L = 1 – f_s$$. The dynamic viscosity $$\mu$$ (Pa·s) of the liquid metal was assumed to follow an Arrhenius-type relation:
$$
\mu(T) = \mu_0 \exp\left(\frac{E_a}{RT}\right)
$$
with $$\mu_0 = 1.2 \times 10^{-4}$$ Pa·s and activation energy $$E_a = 30$$ kJ/mol.
Boundary Conditions and Simulation Setup
I imported the 3D models of the three investment casting designs (including the mold shell) from CAD software into ProCAST. The shell mold was made of a ceramic material with a thickness of 8 mm and thermal conductivity $$k_{\mathrm{mold}} = 0.8$$ W/(m·K). The initial temperature of the mold was set to 300 °C, and the ambient temperature was 25 °C. The pouring process was simulated with a constant inlet velocity corresponding to a fill time of 15 seconds. The heat transfer coefficient at the casting-mold interface was set to 500 W/(m²·K) for the first 30 seconds and then reduced to 200 W/(m²·K) after complete solidification. Gravity was applied in the negative Z direction. I used the Niyama criterion to predict shrinkage porosity:
$$
N_y = G / \sqrt{R}
$$
where $$G$$ is the thermal gradient (K/mm) and $$R$$ is the cooling rate (K/s). A threshold $$N_y < 1$$ indicates a high probability of micro-porosity. The simulations ran on a multi-core workstation with approximately 500,000 elements per model.
Results and Discussion
Mold Filling Analysis
The mold filling patterns for the three schemes are illustrated in Table 3, which summarizes key observations. In Scheme 1, the liquid metal entered from the bottom of the mold, first filling the left side (large flange region) and then flowing downward to meet the rising metal from the bottom inlet. This caused a merging front that could lead to gas entrapment and oxide films. In Scheme 2, the metal rose smoothly from the bottom without any falling flow or merging turbulence. The entire filling process was stable, with a continuous liquid front moving upward. Scheme 3 showed initial filling of the bottom region, but later the top gate introduced additional metal that fell back and merged with the bottom stream, similar to Scheme 1 but less severe. The mold filling results suggest that Scheme 2 provides the most controlled and quiescent filling, minimizing the risk of entrainment defects.
| Scheme | Filling Pattern | Turbulence Level | Potential Defects |
|---|---|---|---|
| 1 | Falling flow + merging fronts | High | Gas entrapment, oxide films |
| 2 | Smooth bottom-up rise | Low | Minimal |
| 3 | Falling flow from top gate | Moderate | Possible inclusions |
Isolated Liquid Regions Analysis
Solidification simulation revealed the presence of isolated liquid pools (ILP) for each scheme, as summarized in Table 4. An isolated liquid region is a volume of molten metal that becomes separated from the riser supply during solidification, leading to shrinkage porosity if it solidifies last. In Scheme 1, despite the presence of three risers, the thick flange areas were fed well, and no significant ILP formed in the casting. The risers effectively directed the solidification front from the thin walls toward the risers. Scheme 2 also showed no isolated liquid regions throughout the entire casting, thanks to the optimized placement of the side and top risers. The temperature gradient was positive from the thin sections toward the risers. In Scheme 3, however, a small isolated liquid pool was detected near the bottom of the small flange region. This area was far from both risers and had a thicker local section. The chill block helped cool the large flange, but the small flange remained a hot spot. As a result, that region would likely develop micro-porosity.
| Scheme | Presence of ILP | Location | Volume (cm³) |
|---|---|---|---|
| 1 | No | — | 0 |
| 2 | No | — | 0 |
| 3 | Yes | Small flange bottom | 0.15 |
Shrinkage Porosity Prediction
Using the Niyama criterion with a threshold of 2% porosity probability, I predicted the volume and distribution of shrinkage porosity in each investment casting scheme. Table 5 lists the results. Scheme 1 had the largest porosity volume (0.92 cm³), concentrated in the central boss and near the large flange region. This is surprising because Scheme 1 had the most risers. However, the long riser necks and large distance may have caused insufficient feeding. Scheme 3 had a porosity volume of 0.88 cm³, also mainly in the small flange area and a secondary spot near the volute curve. In contrast, Scheme 2 showed negligible porosity, only 0.04 cm³, scattered in non-critical thin areas. The side riser in Scheme 2 provided excellent directional solidification, allowing the thick large flange to be fed effectively. The top riser adequately compensated the small flange. Therefore, Scheme 2 is the optimal design among the three.
| Scheme | Total Porosity Volume (cm³) | Main Locations |
|---|---|---|
| 1 | 0.92 | Central boss, large flange |
| 2 | 0.04 | Minimal, thin edges |
| 3 | 0.88 | Small flange bottom, volute curve |
The superiority of Scheme 2 can be attributed to the following factors. First, the side riser on the large flange directly contacts the thickest section, creating a steep thermal gradient and continuous liquid feeding path. Second, the top riser on the small flange is positioned at the highest point of the casting, ensuring that the last solidifying liquid is contained within the riser. Third, the overall mold height is lower in Scheme 2, which reduces the static pressure and promotes smoother filling. In contrast, Scheme 1’s multiple risers led to complex thermal interference, and the long riser necks became bottlenecks. Scheme 3’s chill block over-cooled the large flange, shifting the hot spot to the small flange without adequate riser compensation.
Experimental Verification
Based on the simulation results, I selected Scheme 2 as the final investment casting process. I then produced a batch of six turbine volute castings using the same gating and riser system, with pouring temperature of 710 °C, mold temperature of 300 °C, and pouring time of 15 seconds. The shell mold was made of a standard silica-based investment ceramic. After casting, the parts were cleaned, and the risers were cut off. I performed DR (digital radiography) inspection on all six castings to evaluate internal quality. The DR images showed no detectable shrinkage cavities, porosity, or inclusions in any of the castings. The echo signals were uniform, indicating sound metal. The surface finish was excellent, and the dimensional tolerance met the specification. This excellent consistency between simulation prediction and experimental outcome validates the accuracy of ProCAST in modeling the investment casting process for this complex AlSiCuSc alloy part.
Figure above shows a representative example of the investment casting setup and a finished turbine volute. The image illustrates the wax pattern, shell mold, and the final metal part after dewaxing and pouring. The successful production of defect-free castings in multiple batches proves that the optimized Scheme 2 is robust and reliable for mass production.
Conclusions
In this study, I conducted a comprehensive numerical simulation and experimental validation of the investment casting process for an AlSiCuSc alloy turbine volute. The key findings are summarized as follows:
- The complex geometry of the turbine volute, with significant wall thickness variations, requires careful design of the gating and riser system to ensure directional solidification and avoid shrinkage porosity.
- Among the three proposed investment casting schemes, Scheme 2—featuring a side riser on the large flange and a top riser on the small flange—produced the best filling behavior, no isolated liquid pools, and the lowest predicted porosity (0.04 cm³).
- Experimental casting trials using Scheme 2 resulted in six sound parts with no internal defects as verified by DR inspection, confirming the predictive capability of ProCAST simulation.
- The use of numerical simulation significantly reduced the need for physical trial-and-error, shortening the development cycle and reducing costs for investment casting of high-integrity components.
This work demonstrates that investment casting, when guided by accurate numerical simulation, can produce high-quality AlSiCuSc alloy turbine volutes suitable for demanding turbocharger applications. The methodology can be extended to other complex thin-walled castings where internal quality is critical.