In my research, I focused on the investment casting process of a K4222 superalloy thin-walled component, specifically an aeroengine pre-swirl nozzle. This component has a maximum outer diameter of 365.73 mm, a minimum wall thickness of 1.53 mm, and a weight of approximately 7.21 kg. The complex geometry, combined with the wide solidification range of K4222, makes the process highly sensitive to casting defects such as shrinkage porosity, misrun, and hot tearing. I systematically combined thermodynamic calculations, numerical simulation, and experimental validation to develop a reliable casting route that minimizes casting defects while meeting stringent quality requirements.
K4222 is a nickel-based precipitation-hardened cast superalloy, corresponding to GTD222, and is widely used for guide vanes and static components in aeroengines. It offers moderate high-temperature strength, excellent elongation, fatigue resistance, oxidation resistance, and corrosion resistance up to about 1000°C. The alloy is strengthened by the γ′ phase (Ni₃(Al,Ti)), and its broad melting temperature range (1292°C–1344°C) makes it prone to micro-shrinkage and composition segregation during solidification. Because the pre-swirl nozzle has many thin-walled curved surfaces, abrupt thickness changes, and large planar areas, the formation of casting defects is particularly challenging to control.
In this study, I designed an initial gating and risering system based on the component geometry and K4222 alloy properties. I then used JMatPro and ProCAST to calculate and compare the thermophysical properties of the alloy. I selected the most accurate dataset for simulation. Through multi-iteration numerical simulations, I investigated the effects of pouring temperature, shell temperature, and pouring time on the evolution of casting defects. Finally, I optimized the gating/riser system, produced trial castings, and conducted comprehensive quality inspections including fluorescent penetrant testing, X-ray radiography, blue-light scanning, and mechanical property tests.
Materials and Methods
The chemical composition of K4222 used in my research is listed in Table 1.
| Ni | C | Cr | Co | W | Al | Ti | Nb | B | Ta | Zr | O |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Bal. | 0.1 | 22.73 | 18.95 | 1.9 | 1.19 | 2.35 | 0.86 | 0.0053 | 1.06 | 0.012 | 0.0013 |
The alloy was melted in a vacuum induction furnace and poured under high vacuum into a ceramic shell preheated to a defined temperature. I used a top-gated gravity pouring system to facilitate directional solidification. The overall experimental flow consisted of:
- Geometrical analysis of the casting and design of initial gating/riser system.
- Calculation of material thermophysical properties using JMatPro and ProCAST.
- Numerical simulation of mold filling and solidification using the finite element method.
- Optimization of pouring parameters and gating system design.
- Trial production and destructive/non-destructive evaluation of the castings.
- Heat treatment and mechanical property verification.
To predict casting defects accurately, I established mathematical models for flow, heat transfer, and solidification. The governing equations are summarized below.
Continuity equation (incompressible flow):
$$
\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y} + \frac{\partial w}{\partial z} = 0
$$
Momentum conservation (Navier-Stokes):
$$
\rho \left( \frac{\partial u}{\partial t} + u\frac{\partial u}{\partial x} + v\frac{\partial u}{\partial y} + w\frac{\partial u}{\partial z} \right) = -\frac{\partial P}{\partial x} + \rho g_x + \mu \nabla^2 u
$$
Energy conservation:
$$
\rho c_p \left( \frac{\partial T}{\partial t} + u\frac{\partial T}{\partial x} + v\frac{\partial T}{\partial y} + w\frac{\partial T}{\partial z} \right) = \frac{\partial}{\partial x}\left(k\frac{\partial T}{\partial x}\right) + \frac{\partial}{\partial y}\left(k\frac{\partial T}{\partial y}\right) + \frac{\partial}{\partial z}\left(k\frac{\partial T}{\partial z}\right) + S
$$
For the prediction of shrinkage porosity, I used the Niyama criterion:
$$
G = \left[ \left( \frac{\partial T}{\partial x} \right)^2 + \left( \frac{\partial T}{\partial y} \right)^2 + \left( \frac{\partial T}{\partial z} \right)^2 \right]^{1/2}
$$
The temperature gradient \(G\) and cooling rate \(R\) were post-processed with a threshold value to identify regions susceptible to micro-shrinkage. This helped me to locate potential casting defects before physical trials.
Thermophysical Property Calculation
Since the accuracy of the thermophysical properties strongly affects the simulation of casting defects, I calculated the relevant properties using two independent approaches: the JMatPro phase diagram calculation package and the ProCAST property database, and compared them with values from a traditional superalloy handbook. I evaluated the thermal conductivity, density, specific heat, latent heat, solid fraction, and liquidus/solidus temperatures.
The thermal conductivity of K4222 increases with temperature, but shows a change in slope near the solidus. The density decreases linearly with temperature in the solid state and drops more sharply upon melting. The specific heat exhibits a peak near the solidus due to phase transformation. I determined the liquidus and solidus temperatures to be approximately 1344°C and 1292°C, respectively. These values were used to set the pouring temperature range.
To simplify the finite element calculation while maintaining accuracy, I selected a representative set of material data points at 100°C intervals from 100°C to 1500°C. The fitted curves for density, thermal conductivity, and specific heat are summarized in Table 2.
| Temperature (°C) | Density (kg/m³) | Thermal conductivity (W/m·K) | Specific heat (J/kg·K) |
|---|---|---|---|
| 100 | 8359 | 12.1 | 410 |
| 300 | 8240 | 16.8 | 450 |
| 600 | 8120 | 21.5 | 510 |
| 900 | 7980 | 25.9 | 600 |
| 1200 | 7740 | 28.8 | 720 |
| 1500 | 7340 | 32.0 | 780 |
I verified the calculated solidification range using the Ni-Cr binary phase diagram. The equilibrium liquidus is near 1350°C, which agrees reasonably well with the measured data. The presence of other elements such as Co, W, Ti, Nb, and Ta modifies the solidification path and enlarges the mushy zone, increasing the risk of porosity-type casting defects. Consequently, I paid special attention to the feeding design and the selection of pouring temperature.
Initial Casting Process Design
Based on the component geometry and the alloy characteristics, I set the initial process parameters as follows:
- Pouring temperature range: 1400°C–1550°C
- Shell preheating temperature range: 1050°C–1200°C
- Pouring time range: 3.5 s–5.0 s
- Vacuum degree: < 3 Pa
- Mold material: ceramic shell (fused silica + alumina)
- Interface heat transfer coefficient: 1000 W·m⁻²·K⁻¹
The pouring temperature was selected to be 50°C–200°C above the liquidus temperature. Since the casting is a large thin-walled structure, I chose a top-gating system. The top-gate approach enables good feeding from the risers, but requires careful design to avoid splashing and gas entrapment. I designed the initial gating system with a central sprue of 100 mm diameter, nine runners of 40 mm diameter, and multiple rectangular risers placed at the flanges. The riser dimensions were calculated using the modulus method:
$$
M = \frac{V}{A}
$$
where \(V\) is the volume and \(A\) is the cooling surface area of the riser/casting region. For a riser to feed a casting region, the modulus of the riser must be larger than that of the casting region. In my initial design, I calculated the required riser heights and diameters for the upper and lower flanges, and positioned them symmetrically around the circumference.
The pouring time was estimated using the empirical formula:
$$
t = S \sqrt[3]{G}
$$
where \(t\) is pouring time (s), \(G\) is the casting mass (kg), and \(S\) is an empirical coefficient ranging from 3 to 4.3 for thin-walled castings. For \(G = 7.21\) kg, I obtained \(t = 3.5\)–5.0 s. The corresponding liquid metal velocity in the mold cavity was:
$$
v = \frac{H}{t}
$$
Given a total pouring height \(H\) of about 480 mm, the velocity was 96–137 mm/s, suitable for thin sections below 4 mm in thickness. These parameters were subsequently used as initial conditions for the ProCAST simulation.
Numerical Simulation of Thin-Wall Casting Defects
I first modeled a simplified 100 mm × 100 mm × 1 mm flat plate with the same gating ratio to investigate the influence of process parameters on the formation of casting defects. The mesh was generated with a non-uniform strategy: 1 mm elements for the plate, 5 mm for the ingate, and 8 mm for the sprue and pouring cup. The shell thickness was 6 mm. The simulations were performed using ProCAST with the finite-element method.
Effect of Pouring Temperature
I selected four pouring temperatures – 1400°C, 1450°C, 1500°C, and 1550°C – while keeping the shell temperature at 1050°C, the pouring time at 4 s, and the heat transfer coefficient at 1000 W·m⁻²·K⁻¹. The temperature fields at 35% filling showed that at 1400°C, the metal front cooled rapidly, forming a concentric ring pattern and leading to poor fluidity. At higher pouring temperatures, the temperature distribution became more uniform, and the fluidity improved. At 1500°C and above, the filling capability became nearly saturated.
The shrinkage porosity predictions (Figure omitted) indicated that the largest casting defects appeared at the interface between the thin plate and the ingate because the ingate solidified earlier the plate due to its larger volume. This problem was partly alleviated by increasing the pouring temperature, which extended the solidification time and allowed better feeding. However, at 1550°C, the cooling time became excessively long, and the central region of the plate again showed micro-shrinkage due to slow cooling and a wide mushy zone. The pore fraction variation with temperature is shown in Table 3.
| Pouring temperature (°C) | Predicted porosity (%) | Filling quality |
|---|---|---|
| 1400 | 2.8 | Poor, misrun risk |
| 1450 | 1.9 | Fair |
| 1500 | 1.2 | Good |
| 1550 | 1.7 | Overheating, larger grains |
Therefore, I chose 1500°C as the optimal pouring temperature for subsequent simulations.
Effect of Shell Temperature
I varied the shell temperature from 1050°C to 1200°C in steps of 50°C while fixing the pouring temperature at 1500°C and the pouring time at 4 s. Increasing the shell temperature improves the thermal insulation of the mold, which helps the liquid metal flow into thin sections. However, if the shell temperature is too high, the cooling rate of the casting decreases, which promotes the formation of shrinkage porosity in the center of thick sections. The simulation results are summarized in Table 4.
| Shell temperature (°C) | Predicted porosity (%) | Observed tendency |
|---|---|---|
| 1050 | 1.2 | Good filling, low porosity |
| 1100 | 1.4 | Better fluidity, still low porosity |
| 1150 | 1.9 | Increasing centerline shrinkage |
| 1200 | 2.5 | High porosity, risk of mold-metal reaction |
I selected 1100°C as the optimum shell temperature because it offers a good balance between filling capacity and acceptable casting defect levels. At 1200°C, the increased porosity is not acceptable for aerospace parts.
Effect of Pouring Time
I then varied the pouring time from 3.5 s to 5.0 s while maintaining the pouring temperature at 1500°C and the shell temperature at 1100°C. The pouring time controls the average filling velocity. A shorter pouring time (higher filling rate) reduces the time for the liquid to cool inside the mold, thus lowering the risk of misrun. However, too high a pouring rate can cause erosion, splashing, and gas entrapment. The simulations showed that the temperature field at the end of filling becomes more uniform as the pouring time decreases. The optimum pouring time was found to be 4.5 s, giving a predicted porosity of about 1.1%. At 5.0 s, some thin sections showed incomplete filling, while at 3.5 s, the risk of inclusion increased. Therefore, the final process parameters were set as shown in Table 5.
| Parameter | Value |
|---|---|
| Pouring temperature | 1500°C |
| Shell temperature | 1100°C |
| Pouring time | 4.5 s |
| Vacuum degree | < 3 Pa |
| Heat transfer coefficient | 1000 W/m²K |
Gating and Riser System Optimization
I first simulated the casting with the initial gating system. The shrinkage porosity prediction of the bare casting (without gates) showed that the largest casting defects were located at the thick flanges, which are the natural hot spots. The risers in the initial system were placed at the top and bottom flanges, but they were “cold risers” because they were filled with metal that had already passed through the casting. This limited their feeding efficiency. The predicted defect distribution indicated significant porosity in the lower flange and in the junction areas.
To overcome this issue, I modified the gating system by adding an internal runner system connected directly to the lower flange risers. In the optimized design, the sprue feeds both the casting through the upper ingates and the bottom risers through internal runners. This allows the bottom risers to be filled with hot metal directly from the sprue, which greatly improves the liquid metal feeding. The optimized gating system is shown schematically in the following figure.
The optimized system retained the central sprue of 100 mm diameter but included nine internal runners connecting the sprue to the lower flange risers. The upper flange remained connected to the direct pour cup. With this design, the mold filling process became more uniform. The simulated filling sequence showed that the metal front advanced smoothly from the top and the bottom toward the center, with no obvious turbulence or splashing. The solidification sequence proceeded from the thin walls toward the thick flanges, with the risers solidifying last, thus providing adequate feeding.
The shrinkage porosity predictions for the optimized system (threshold 2%) showed that most of the casting defects were confined to the riser areas and could be removed during machining. The internal porosity in the thin walls was greatly reduced compared to the initial design. Figure 1 below shows the comparison of defect volume at different casting locations between the initial and optimized gating systems.
| Location | Initial system (cm³) | Optimized system (cm³) |
|---|---|---|
| Top flange | 2.4 | 0.8 |
| Thin walls | 1.5 | 0.3 |
| Bottom flange | 4.6 | 0.9 |
| Inner junctions | 1.2 | 0.4 |
The optimized gating system successfully minimized casting defects within the body of the pre-swirl nozzle. The simulation time to full solidification was about 2.5 hours, which is acceptable for a thin-walled superalloy casting of this size.
Experimental Validation and Quality Inspection
I produced trial castings using the optimized gating system and the process parameters listed in Table 5. After casting, the shells were knocked out and the castings were cut off from the gating system. The castings were then subject to heat treatment according to the following schedule:
- Solution treatment: 1150°C for 4 hours, air cool.
- Aging treatment: 800°C for 8 hours, air cool.
Surface Quality and Fluorescent Penetrant Testing
The visual inspection of the castings showed no sand inclusion, cold shut, or surface cracks. The surface was smooth and free of flash. Following the standard procedure for fluorescent penetrant inspection, I applied a Type I Method A penetrant (PSM-5) and a developer (ZP-48) with a penetration time of 15 minutes. Under black light (≥1200 μW/cm²), the castings showed no linear indications. According to the standard QC-57.220255 Rev.A, no surface casting defects larger than 0.3 mm were observed. This confirmed that the optimized process prevents surface-related casting defects.
X-ray Inspection
I performed X-ray radiography on the entire casting using an XYD3010 machine with N-Ⅲ film. The parameters were: exposure time 3 min, focus-to-film distance 1000 mm, tube current 5 mA, and tube voltage 80 kV. The evaluation was performed according to ASTM E 192 reference radiographs. The X-ray images showed only minor scattered micro-porosity in the thin walls, all below the acceptance limit of 0.3 mm. The casting defects detected were within the allowable limits and, in many cases, were located in zones that would be removed by subsequent machining. The agreement between the simulated defect locations and the X-ray inspection results was good, thereby confirming the reliability of the simulation.
Dimensional Accuracy via Blue-light Scanning
To evaluate the dimensional accuracy of the cast pre-swirl nozzle, I used a StereoScan neo R8 blue-light scanner with 8 megapixel cameras and a minimum point spacing of 17 μm. I scanned the entire casting and compared the obtained point cloud with the original CAD model using Imageware software. I set the critical tolerance was set to ±0.75 mm. The results showed that 98.12% of the measured points fell within this tolerance. The maximum deformation observed was about 0.755 mm, which is within the acceptable range for this component. The blue-light scan confirmed that the investment casting process, combined with the optimized gating system, produced a dimensionally accurate component with minimal distortion.
Mechanical Property Verification
Since the pre-swirl nozzle itself cannot be easily machined into tensile specimens due to its complex geometry, I cast test bars integrally with the same gating system and heat treated them together with the casting. The tensile properties were measured at room temperature and at 760°C. The results are shown in Tables 7 and 8.
| Property | Requirement | Measured |
|---|---|---|
| Yield strength (MPa) | ≥ 724 | 755 |
| Tensile strength (MPa) | ≥ 896 | 980 |
| Elongation (%) | ≥ 4 | 11 |
| Reduction of area (%) | ≥ 6 | 14 |
| Property | Requirement | Measured |
|---|---|---|
| Yield strength (MPa) | ≥ 552 | 661.5 |
| Tensile strength (MPa) | ≥ 703 | 840.0 |
| Elongation (%) | ≥ 6 | 10.0 |
| Reduction of area (%) | ≥ 8 | 17.5 |
I also performed stress-rupture tests at 900°C under an initial stress of 172 MPa. The requirement was a rupture life of at least 30 hours and an elongation of at least 7%. The measured test bar broke after 47.4 hours, with an elongation of 28.5% and a reduction of area of 29.0%. All mechanical properties satisfy the technical requirements of the pre-swirl nozzle, indicating that the optimized casting process not only suppresses casting defects but also produces sound material with the desired mechanical performance.
Conclusion
In this work, I developed and optimized an investment casting process for a K4222 superalloy pre-swirl nozzle with emphasis on minimizing casting defects. The key findings and achievements are as follows:
- I analyzed the complex geometry of the pre-swirl nozzle and identified the main challenges: large thin-walled areas, abrupt thickness changes, long feeding distances, and the wide solidification range of K4222. These features require careful control of the gating system and process parameters to avoid casting defects.
- I calculated the thermophysical properties of K4222 using JMatPro and ProCAST and selected the most appropriate dataset for simulation. The liquidus and solidus temperatures were found to be 1344°C and 1292°C, respectively.
- Through systematic numerical simulations of a thin-walled plate with different pouring temperatures, shell temperatures, and pouring times, I determined the optimal processing window: pouring temperature 1500°C, shell temperature 1100°C, and pouring time 4.5 s. These parameters minimize the risk of shrinkage porosity and ensure complete filling.
- The initial gating system with cold risers was insufficient to provide adequate feeding. I optimized the system by adding internal runners from the sprue to the bottom flange risers. The optimized design significantly reduced casting defects in the casting body, as confirmed by simulations.
- Experimental trial castings were produced using the optimized process. Fluorescent penetrant testing and X-ray radiography confirmed the absence of unacceptable casting defects. Blue-light scanning showed that the dimensional accuracy stayed within ±0.75 mm for 98.12% of the surface points, with a maximum deviation of about 0.755 mm.
- Heat-treated test bars exhibited excellent room-temperature and high-temperature tensile properties, as well as stress-rupture properties, meeting or exceeding the technical requirements for the pre-swirl nozzle.
The combination of material modeling, casting simulation, and experimental validation proved to be a powerful approach for developing a robust investment casting process for complex thin-walled superalloy components. This methodology not only reduces development time and cost but also improves the quality and reliability of aerospace castings by systematically minimizing casting defects.
In the future, I plan to extend this work to other K4222 components with even thinner walls and more complex internal cooling passages. Further studies could include the effect of mold materials and coating layers on interfacial heat transfer, as well as the prediction of residual stresses and deformation using coupled thermo-mechanical simulations. The ultimate goal is to establish a fully digitalized process chain for investment casting of aerospace superalloys that ensures defect-free production and consistent quality.