In the development of solid blades for a heavy-duty gas turbine, precision investment casting is a critical manufacturing process due to its ability to produce complex geometries with high dimensional accuracy. However, controlling dimensions throughout the precision investment casting process poses significant challenges, particularly for large components like turbine blades, where non-uniform shrinkage and deformation can lead to out-of-tolerance parts. This article details a comprehensive approach to dimension control in precision investment casting, focusing on a blade approximately 400 mm in length and 3 kg in mass, made from a high-temperature alloy. The methodology integrates advanced techniques such as 3D printing for process development, MoldFlow simulation for wax pattern optimization, and digital scanning for dimensional inspection. By systematically addressing each stage—from wax pattern fabrication to shell sintering and final casting—we achieved blade dimensions within a tolerance of ±0.30 mm, demonstrating the effectiveness of the proposed strategies in precision investment casting.
The precision investment casting process involves multiple steps: wax pattern creation, shell building, dewaxing, sintering, pouring, and post-processing. Each step introduces potential dimensional variations due to material shrinkage, thermal expansion, and mechanical stresses. For the turbine blade in question, the primary goal was to minimize these variations to meet stringent aerodynamic and structural requirements. We began by using 3D-printed photopolymer resin models to experimentally determine the overall casting shrinkage factor, which is essential for designing the wax injection mold. The resin models were coated with a low-melting-point wax to prevent shell cracking during dewaxing, and key measurement marks were inscribed for dimensional tracking. After casting, the shrinkage from resin model to final casting was calculated, yielding values of approximately 1.13% in the length direction and 1.45% in the width and thickness directions. Combining these with the wax material’s free shrinkage rate of 0.60%, the mold shrinkage rates were derived, as summarized in Table 1.
| Direction | Shrinkage Rate (%) |
|---|---|
| Width (X) | 2.05 |
| Thickness (Y) | 2.05 |
| Length (Z) | 1.73 |
These shrinkage rates were applied to the CAD model of the blade to generate the wax injection mold design. The mold was then used to produce wax patterns, but initial trials revealed excessive deformation, especially in thick sections like the blade root. To address this, we employed MoldFlow software to simulate the wax injection and cooling processes. The simulation used a 3D mesh with 19,745 elements and 9,762 nodes, modeling both solid wax patterns and patterns with a cold wax core. The cold wax core technique involves inserting a pre-cooled wax insert into the mold before injection, which reduces thermal gradients and promotes uniform cooling. The simulation parameters included typical injection settings for similar blades, with cooling channels arranged as shown in the mold design. The results indicated that without a cold wax core, the volumetric shrinkage varied significantly across the blade: 5.18% at the thick root section and 2.34% at the thin trailing edge, leading to a differential of 2.51%. With the cold wax core, shrinkage was reduced to 3.37% at the root and 2.76% at the maximum thickness section, with a differential of only 0.41%. This confirms that the cold wax core effectively minimizes non-uniform shrinkage in precision investment casting, as expressed by the shrinkage reduction factor $$ \Delta S = S_{\text{without core}} – S_{\text{with core}} $$ where \( S \) represents volumetric shrinkage. For instance, at the root, \( \Delta S = 5.18\% – 3.37\% = 1.81\% \), indicating a 35% improvement.
Following the simulation, actual wax patterns were produced using F28-44B medium-temperature wax. The patterns were inspected with an ATOs blue light scanner, which has an accuracy of 0.02 mm. The dimensional data showed that patterns with the cold wax core had a maximum deviation of -0.25 mm in the blade body, whereas those without the core deviated by up to -0.56 mm, exceeding tolerance limits. Additionally, infrared thermography with a Fluke Ti29 camera was used to analyze temperature distributions during cooling. Patterns without the cold wax core exhibited temperatures ranging from 23.3°C to 48.5°C, with an average of 36.4°C, leading to uneven cooling and deformation. In contrast, patterns with the cold wax core had a more uniform temperature range of 23.3°C to 41.6°C, averaging 27.5°C, which contributed to consistent shrinkage. The relationship between linear shrinkage and wax thickness was also investigated, as shown in Figure 1, indicating that shrinkage increases with thickness, from 1.15% at 2.5 mm to 1.40% at 40 mm. This underscores the importance of controlling wall thickness in precision investment casting to manage dimensional stability.
To further mitigate deformation, wax patterns were placed on custom-made fixtures immediately after ejection from the mold. These fixtures, shaped to match the blade’s concave or convex surfaces, constrained the patterns during cooling, preventing twisting and warpage. MoldFlow’s warpage analysis principles highlight that differential cooling between surfaces causes asymmetric shrinkage, leading to bending. The fixture ensured synchronous cooling, reducing deviations to within -0.20 mm to 0.10 mm, compared to free-cooled patterns that showed significant twisting and shrinkage. This step is crucial in precision investment casting for maintaining dimensional integrity before shell building.
The next phase involved shell construction, which can introduce additional dimensional changes due to ceramic slurry application, drying, and sintering. The shell was built using an 8-layer process: a primary coat of 320-mesh white alumina powder in silica sol, stuccoed with 80-mesh white alumina sand; intermediate coats with the same slurry but stuccoed with 46-mesh sand; and backup coats with 24-mesh sand, followed by a seal coat. After dewaxing and sintering at 1100°C, the shell was dissected to inspect its internal dimensions. ATOs scanning revealed minimal deformation on the concave side, with deviations within ±0.10 mm, while the convex side near the leading edge at the tip showed a contraction of 0.20 mm. This provides valuable insight into shell behavior in precision investment casting, often treated as a “black box.” The shell’s dimensional stability can be modeled using thermal expansion coefficients, where the total dimensional change \( \Delta L \) during sintering is given by $$ \Delta L = L_0 \cdot \alpha \cdot \Delta T $$ where \( L_0 \) is the initial length, \( \alpha \) is the thermal expansion coefficient of the ceramic, and \( \Delta T \) is the temperature change. For the alumina-based shell, \( \alpha \approx 8 \times 10^{-6} /°C \), leading to negligible expansion, but shrinkage from binder removal can occur.
After shell preparation, casting was performed by preheating the shell to 1100°C and pouring the alloy at 1490°C. The cast blades were cleaned, shot-peened, and scanned to evaluate final dimensions. The overall shrinkage from mold to casting was calculated as 1.64% in the length direction, closely matching the design value of 1.70%, validating the mold shrinkage settings. When comparing the cast blade to the theoretical model via best-fit alignment, dimensional deviations across the blade body were within ±0.30 mm, meeting specifications. This success is attributed to the integrated approach in precision investment casting, combining empirical data from 3D-printed models, simulation-driven wax pattern optimization, and rigorous inspection.
To delve deeper into the factors affecting dimension control in precision investment casting, we can analyze the contribution of each process variable. The total dimensional error \( E_{\text{total}} \) in a cast blade can be expressed as a sum of errors from individual stages: $$ E_{\text{total}} = E_{\text{wax}} + E_{\text{shell}} + E_{\text{casting}} + E_{\text{thermal}} $$ where \( E_{\text{wax}} \) is wax pattern error, \( E_{\text{shell}} \) is shell deformation error, \( E_{\text{casting}} \) is solidification shrinkage error, and \( E_{\text{thermal}} \) is thermal contraction error. Our study shows that \( E_{\text{wax}} \) is dominant, reduced by the cold wax core and fixtures, while \( E_{\text{shell}} \) is minimal due to controlled sintering. Additionally, the alloy’s solidification shrinkage, typically around 2-3% for high-temperature alloys, was compensated by the mold design. The effectiveness of the cold wax core can be quantified by the uniformity index \( U \), defined as $$ U = 1 – \frac{\sigma_{\text{shrinkage}}}{\mu_{\text{shrinkage}}} $$ where \( \sigma_{\text{shrinkage}} \) is the standard deviation of shrinkage across the blade and \( \mu_{\text{shrinkage}} \) is the mean shrinkage. For patterns without the core, \( U \) was low (indicating high non-uniformity), but with the core, \( U \) approached 1, signifying uniform shrinkage. This metric is useful for optimizing precision investment casting processes for complex geometries.
Further considerations include the impact of wax material properties on dimension control. The wax used, F28-44B, has a specific thermal conductivity \( k \) and coefficient of thermal expansion \( \beta \), which influence cooling rates and shrinkage. The cooling time \( t_c \) for a wax pattern can be estimated using Fourier’s law: $$ t_c = \frac{\rho c_p d^2}{k} $$ where \( \rho \) is density, \( c_p \) is specific heat, and \( d \) is thickness. Thicker sections cool slower, leading to higher shrinkage, as observed. By using a cold wax core, the effective thickness \( d \) is reduced, decreasing \( t_c \) and promoting uniform cooling. This principle is fundamental to precision investment casting for heavy sections. Additionally, the wax injection parameters—such as pressure, temperature, and holding time—were optimized based on simulation results to minimize residual stresses. Table 2 summarizes key parameters and their effects on dimensional accuracy.
| Parameter | Optimal Value | Effect on Dimension Control |
|---|---|---|
| Injection Temperature | 70°C | Reduces viscosity for better fill, minimizes thermal gradients |
| Cooling Time | 180 s | Ensures uniform solidification, prevents warpage |
| Cold Wax Core Temperature | 15°C | Enhances heat extraction, reduces shrinkage variation |
| Fixture Contact Time | 30 min | Constrains deformation during critical cooling phase |
The shell-building process also plays a role in dimension control. The ceramic shell must have sufficient strength to withstand handling and sintering while minimizing distortion. The slurry viscosity and stucco size affect shell thickness and uniformity, which in turn influence thermal mass during sintering. We measured shell thickness variations using cross-sectional analysis and found them to be within 0.5 mm, contributing to consistent heating and cooling. The sintering process involves phase transformations in the binder, which can cause shrinkage. By pre-sintering at 1100°C, the shell stabilizes, reducing subsequent dimensional changes during casting. This step is often overlooked in precision investment casting but is critical for high-tolerance parts.
In terms of metallurgical aspects, the alloy used (similar to 4716) has a known solidification range and shrinkage behavior. The pouring temperature of 1490°C was selected to ensure fluidity while minimizing thermal shock to the shell. The solidification shrinkage \( \epsilon_s \) can be calculated as $$ \epsilon_s = \frac{\rho_l – \rho_s}{\rho_l} $$ where \( \rho_l \) and \( \rho_s \) are the liquid and solid densities, respectively. For typical nickel-based superalloys, \( \epsilon_s \) is around 2-3%, which aligns with our observed casting shrinkage. By incorporating this into the mold design, we achieved accurate final dimensions. Furthermore, post-casting heat treatments were applied to relieve stresses and stabilize microstructure, but their effect on dimensions was minimal due to controlled cooling rates.
The use of digital scanning technologies, such as the ATOs blue light scanner, was instrumental throughout the precision investment casting process. It allowed for non-contact, high-precision measurement of wax patterns, shells, and castings, enabling rapid feedback and correction. The data collected were used to create deviation maps, highlighting areas prone to error. For instance, the root section consistently showed higher shrinkage, prompting additional cooling in the mold design. This iterative approach, combining simulation and empirical data, is a hallmark of modern precision investment casting for aerospace components.
Looking at broader applications, the methodologies developed here can be extended to other complex parts in precision investment casting, such as turbine vanes or structural components. The cold wax core technique is particularly beneficial for parts with varying cross-sections, while fixture-based cooling can be adapted to different geometries. Moreover, the integration of 3D printing for rapid prototyping reduces development time and cost, as physical molds can be finalized with greater confidence. This aligns with industry trends toward digitalization and additive manufacturing in foundry processes.
In conclusion, dimension control in precision investment casting for heavy-duty gas turbine blades requires a holistic approach that addresses each stage of the process. Through the use of 3D-printed resin models for shrinkage determination, MoldFlow simulation for wax pattern optimization, cold wax core and fixture techniques for deformation reduction, and digital scanning for verification, we achieved blade dimensions within specified tolerances. The key findings emphasize the importance of uniform cooling in wax patterns, controlled shell sintering, and accurate mold design based on empirical shrinkage data. These strategies not only improve dimensional accuracy but also enhance yield and reduce rework, making precision investment casting a reliable method for high-performance components. Future work could explore real-time monitoring of temperature and stress during casting, as well as machine learning algorithms to predict shrinkage patterns, further advancing the state of the art in precision investment casting.