The pursuit of manufacturing high-integrity, complex geometry components for demanding applications, such as turbomachinery, drives the continuous advancement of foundry techniques. Among these, lost wax investment casting stands out as a premier near-net-shape process capable of producing parts with excellent surface finish, dimensional accuracy, and intricate details. However, the inherent complexity of the process—involving mold filling, heat transfer, solidification, and potential defect formation—makes traditional trial-and-error methods costly and time-consuming. This is particularly true for critical components like integral impellers, which often feature thin, twisted blades and thick hubs, creating challenging thermal conditions that can lead to defects such as shrinkage porosity and misruns. In my investigation, I employ numerical simulation as a powerful tool to virtually prototype, analyze, and optimize the lost wax investment casting process for a specific aluminum alloy impeller, thereby demonstrating a systematic approach to enhancing casting quality and manufacturability.
The core challenge in lost wax investment casting of an impeller lies in managing the solidification pattern. The thin blades solidify rapidly, potentially isolating sections of the thicker hub from liquid metal feed, leading to internal shrinkage defects. The design of the gating system, which includes the sprue, runners, and gates, is paramount in controlling the sequence of filling and establishing favorable thermal gradients for directional solidification. To systematically address this, I define two preliminary gating system designs for a gravity-poured process: a top-gating system and a side-gating system. The initial geometry of the impeller and the two system concepts are modeled in CAD software, adhering to the principles of lost wax investment casting to ensure proper wax assembly and de-waxing capability. A comparison of the fundamental gating approaches considered is summarized below.
| Gating Method | Description | General Characteristics |
|---|---|---|
| Top-Gating | Metal enters the mold cavity from the top. | Promotes thermal gradients favorable for directional solidification; higher risk of mold erosion and turbulence. |
| Side-Gating | Metal enters the mold cavity horizontally or at an angle from the side. | Reduces mold冲击; good venting; convenient for wax pattern assembly. |
| Bottom-Gating | Metal enters from the bottom of the mold. | Very smooth filling, excellent slag trapping; can lead to reverse temperature gradients. |
| Combined Gating | A hybrid of the above methods. | Used for large, complex castings with multiple thermal centers. |
The physical phenomena occurring during lost wax investment casting are governed by the fundamental laws of fluid dynamics and heat transfer. To simulate these accurately, a robust mathematical framework is established. The filling stage is treated as a transient, incompressible, viscous fluid flow with a free surface. The governing equations for mass, momentum, and energy conservation are applied.
The continuity equation ensuring mass conservation is:
$$ \frac{\partial \rho}{\partial t} + \frac{\partial (\rho u_x)}{\partial x} + \frac{\partial (\rho u_y)}{\partial y} + \frac{\partial (\rho u_z)}{\partial z} = 0 $$
where $\rho$ is the fluid density and $u_x, u_y, u_z$ are the velocity components.
The momentum conservation (Navier-Stokes) equation is expressed as:
$$ \rho \frac{du_x}{dt} = \rho F_x – \frac{\partial P}{\partial x} + \frac{\partial}{\partial x}\left[\mu\left(2\frac{\partial u_x}{\partial x} – \frac{2}{3}\nabla \cdot \vec{u}\right)\right] + \frac{\partial}{\partial y}\left[\mu\left(\frac{\partial u_x}{\partial y} + \frac{\partial u_y}{\partial x}\right)\right] + \frac{\partial}{\partial z}\left[\mu\left(\frac{\partial u_z}{\partial x} + \frac{\partial u_x}{\partial z}\right)\right] $$
where $P$ is pressure, $\mu$ is dynamic viscosity, and $F_x$ represents body forces.
The energy equation accounting for heat transfer during filling is:
$$ \frac{de}{dt} + p \frac{d}{dt}\left(\frac{1}{\rho}\right) = \frac{1}{\rho} \nabla \cdot (\lambda \nabla T) + \frac{\phi}{\rho} $$
where $e$ is internal energy, $\lambda$ is thermal conductivity, $T$ is temperature, and $\phi$ is the viscous dissipation function.
Following the filling phase, the solidification simulation becomes critical. The primary mode of heat loss is conduction through the metal and into the ceramic shell. The governing heat conduction equation in a cylindrical coordinate system (relevant for rotational parts like impellers) is:
$$ \rho c_p \frac{\partial T}{\partial t} = \frac{1}{r} \cdot \lambda \frac{\partial T}{\partial r} + \frac{\partial}{\partial r}\left(\lambda \frac{\partial T}{\partial r}\right) + \frac{1}{r^2} \cdot \frac{\partial}{\partial \theta}\left(\lambda \frac{\partial T}{\partial \theta}\right) + \frac{\partial}{\partial z}\left(\lambda \frac{\partial T}{\partial z}\right) + \dot{Q} $$
where $c_p$ is the specific heat capacity and $\dot{Q}$ is an internal heat source term (e.g., latent heat of fusion).
The accurate prediction of shrinkage porosity, a key defect in lost wax investment casting, relies on a solidification feeding criterion. I employ the widely used Niyama criterion, which is a local thermal parameter-based model. It postulates that shrinkage porosity is likely to form in regions where:
$$ \frac{G}{\sqrt{\dot{T}}} < C_{\text{Niyama}} $$
Here, $G$ is the local temperature gradient ($^\circ$C/cm), $\dot{T}$ is the local cooling rate ($^\circ$C/s), and $C_{\text{Niyama}}$ is a material-dependent critical value. For many aluminum alloys, a typical critical value is approximately 1 $^\circ\mathrm{C}^{1/2}\mathrm{cm}^{-1}\mathrm{s}^{1/2}$. The gradient $G$ between a cell at temperature $T_{i,j,k}$ and its neighbor is calculated as:
$$ G = \max \left( \frac{T_{i,j,k} – T_{i+m_i, j+m_j, k+m_k}}{\sqrt{(m_i \Delta x)^2 + (m_j \Delta y)^2 + (m_k \Delta z)^2}} \right) $$
where $m_i, m_j, m_k = -1, 0, 1$ and $\Delta x, \Delta y, \Delta z$ are cell dimensions.
To implement this mathematical framework, a comprehensive pre-processing phase is conducted. The 3D CAD models of the two gating system assemblies are imported into the simulation environment. A critical step is mesh generation. To balance computational accuracy and efficiency, a non-uniform meshing strategy is adopted. The impeller itself, with its thin blades and complex curvature, is meshed with a fine element size to capture thermal gradients accurately. The larger volumes of the gating system (sprue, runners, pouring cup) are meshed with a coarser size. After surface meshing and repair, the top-gating and side-gating systems yield 39,104 and 48,472 surface elements, respectively. A 5 mm thick ceramic shell is then added around the entire assembly, and volume meshing generates approximately 317,707 and 567,346 tetrahedral elements for the two systems, respectively. This meticulous meshing is crucial for a reliable lost wax investment casting simulation.
The material properties assigned are central to the model’s fidelity. The impeller is specified as ZAlSi7MgA (ZL101A) aluminum alloy, a common choice for investment castings due to its good castability and mechanical properties. The ceramic mold is defined as fused silica. The key material properties for the alloy are summarized below.
| Property Type | Details |
|---|---|
| Chemical Composition (wt.%) | Si: 7±0.5; Mg: 0.35±0.1; Ti: 0.08~0.20; Bal. Al |
| Thermal Properties | Liquidus: ~610 $^\circ$C; Solidus: ~577 $^\circ$C |
| Mechanical Properties (Typical) | Tensile Strength: $\geq$ 225 MPa; Elongation: $\geq$ 5%; Hardness: $\geq$ 60 HB |
| Density | ~2.68 g/cm$^3$ at 20$^\circ$C; ~2.40 g/cm$^3$ at 700$^\circ$C |
The boundary conditions are defined to reflect the actual lost wax investment casting process. The interface heat transfer coefficient (HTC) between the molten aluminum and the quartz-based shell is a significant parameter. Based on literature and model calibration for similar processes, an HTC value of 1000 W/(m$^2\cdot$K) is applied. The initial process parameters are set to promote sequential solidification: a relatively high mold preheat temperature to slow cooling and a controlled pouring speed. For the top-gating system, initial parameters are a pouring temperature of 720$^\circ$C and a mold preheat of 350$^\circ$C. For the side-gating system, they are 710$^\circ$C and 350$^\circ$C, respectively. The pouring speed is calculated using empirical formulas like the Carlin formula to ensure complete filling:
$$ v_{\text{pour}} = 0.22 \times \frac{\sqrt{h}}{\delta \cdot \ln\left(\frac{t}{380}\right)} $$
where $h$ is the casting height (cm), $\delta$ is the casting wall thickness (cm), and $t$ is the pouring temperature ($^\circ$C). This yielded initial speeds of ~23 mm/s for top-gating and ~35 mm/s for side-gating.
The simulation results for the two initial gating systems provide clear insights. The defect prediction maps, primarily based on the Niyama criterion, reveal significant issues. In the top-gating system, severe shrinkage porosity is predicted in the thin sections of the blades and at the junctions between the blades and the central hub. The side-gating system shows a marked improvement, with a notable reduction in the extent and severity of predicted shrinkage. However, it is not defect-free; residual porosity is still indicated at the top of the hub and in the upper regions of some blades. The side-gating system’s superior performance is attributed to its more controlled, less turbulent filling pattern, which helps establish a more favorable initial temperature distribution. This comparative analysis clearly identifies the side-gating configuration as a more promising baseline for further optimization in this lost wax investment casting application.
Building on the initial results, I focus on optimizing the side-gating system. The analysis indicates that the top of the hub is a last-to-solidify hot spot that is poorly fed because the thin blades solidify early and isolate it from the liquid metal in the gates. The logical optimization is to add a feeding source directly to this region. Therefore, I modify the geometry by incorporating a cylindrical riser (or feeder) on top of the impeller’s hub. This riser is designed to remain molten longer than the hub, providing a reservoir of liquid metal to compensate for the solidification shrinkage occurring in the hub and the adjacent blade roots. The modified 3D model is re-meshed, maintaining the same level of refinement for the impeller.
With the improved geometry, the next step is to refine the process parameters to minimize defects further. I select three key variables for optimization: Pouring Temperature (A), Pouring Speed (B), and Mold Preheat Temperature (C). An orthogonal experimental design L9(3^3) is employed to efficiently explore the multi-parameter space with a limited number of simulations. The factor levels are chosen around the initial baseline values.
| Level | Factor A: Pouring Temp. ($^\circ$C) | Factor B: Pouring Speed (mm/s) | Factor C: Mold Preheat ($^\circ$C) |
|---|---|---|---|
| 1 | 690 | 35 | 320 |
| 2 | 710 | 30 | 350 |
| 3 | 730 | 40 | 380 |
Nine separate numerical simulations are run according to the orthogonal array. The output metric for comparison is the maximum predicted shrinkage porosity percentage in the impeller casting (excluding the riser). The simulation results and analysis are summarized as follows.
| Run No. | A ($^\circ$C) | B (mm/s) | C ($^\circ$C) | Max. Shrinkage (%) |
|---|---|---|---|---|
| 1 | 690 | 35 | 320 | 2.163 |
| 2 | 690 | 30 | 350 | 2.398 |
| 3 | 690 | 40 | 380 | 1.795 |
| 4 | 710 | 35 | 350 | 1.501 |
| 5 | 710 | 30 | 380 | 1.975 |
| 6 | 710 | 40 | 320 | 2.034 |
| 7 | 730 | 35 | 380 | 2.201 |
| 8 | 730 | 30 | 320 | 1.876 |
| 9 | 730 | 40 | 350 | 2.353 |
To analyze the orthogonal results, I calculate the sum of the shrinkage percentage for each factor at each level (K1, K2, K3) and then the range (R) for each factor. A smaller sum indicates better performance for that level, and a larger range indicates a greater influence of that factor on the result.
For Factor A (Pouring Temperature):
K1 (690°C) = 2.163 + 2.398 + 1.795 = 6.356
K2 (710°C) = 1.501 + 1.975 + 2.034 = 5.510
K3 (730°C) = 2.201 + 1.876 + 2.353 = 6.430
Range R_A = max(K1,K2,K3) – min(K1,K2,K3) = 6.430 – 5.510 = 0.920
For Factor B (Pouring Speed):
K1 (35 mm/s) = 2.163 + 1.501 + 2.201 = 5.865
K2 (30 mm/s) = 2.398 + 1.975 + 1.876 = 6.249
K3 (40 mm/s) = 1.795 + 2.034 + 2.353 = 6.182
Range R_B = 6.249 – 5.865 = 0.384
For Factor C (Mold Preheat):
K1 (320°C) = 2.163 + 2.034 + 1.876 = 6.073
K2 (350°C) = 2.398 + 1.501 + 2.353 = 6.252
K3 (380°C) = 1.795 + 1.975 + 2.201 = 5.971
Range R_C = 6.252 – 5.971 = 0.281
The analysis reveals that the optimal combination within the tested levels is A2B1C2, corresponding to a pouring temperature of 710$^\circ$C, a pouring speed of 35 mm/s, and a mold preheat temperature of 350$^\circ$C. This combination (Run 4) yielded the lowest shrinkage porosity prediction of 1.501%. Furthermore, comparing the ranges (R_A > R_B > R_C), it is evident that Pouring Temperature has the most significant influence on the final shrinkage defect in this lost wax investment casting setup, followed by Pouring Speed, and then Mold Preheat Temperature.
A final simulation is run using the optimized geometry (side-gate with top-hub riser) and the optimized process parameters (710$^\circ$C, 35 mm/s, 350$^\circ$C). The result shows a dramatic improvement. The area and severity of predicted shrinkage porosity are confined to a minimal region, predominantly within the safety of the riser itself, with only negligible indications in the uppermost blade roots. The maximum porosity percentage in the casting body is significantly reduced compared to all previous configurations. This virtual outcome strongly suggests a high-quality, sound casting can be achieved. To validate the simulation, a physical casting trial based on this final optimized scheme would be conducted. The expected outcome is an aluminum alloy impeller that is completely filled, with smooth surfaces and no internal shrinkage defects detectable by standard non-destructive testing methods, confirming the effectiveness of the simulation-driven optimization for the lost wax investment casting process.
In conclusion, this work demonstrates a complete, systematic methodology for enhancing the lost wax investment casting of complex thin-walled components. By leveraging numerical simulation, I was able to efficiently evaluate alternative gating designs, identify the superior side-gating approach, diagnose the root cause of residual defects (inadequate feeding at the hub), and implement a geometric solution via a riser. Furthermore, through designed numerical experiments (orthogonal array), I quantitatively optimized the key thermal and flow parameters, identifying their relative influence and determining the optimal set. The final simulation predicts a high-integrity casting. This virtual prototyping approach eliminates costly physical trials, shortens development cycles, and provides deep insight into the solidification behavior, establishing numerical simulation as an indispensable tool in the modern foundry for advancing lost wax investment casting technology.