In the sand casting foundry, the production of large and thick-walled titanium alloy components often faces challenges related to shrinkage porosity and cavity defects. These defects are particularly pronounced in regions with complex geometries and poor heat dissipation, such as junctions between thick and thin sections. Our study focuses on an annular titanium alloy casting produced via sand casting, where we employed numerical simulation to optimize the gating and riser system. The primary objective was to understand the solidification behavior and defect formation mechanisms, and to design a process that eliminates internal shrinkage defects in the casting body. Through systematic simulation and experimental validation, we developed an optimized process that successfully relocates shrinkage cavities into the ingate, ensuring the integrity of the final product.
The casting under investigation has a complex conical annular geometry with an outer diameter of 600 mm, an inner diameter of 370 mm, and a height of 180 mm. The wall thickness varies significantly, ranging from 12 mm to 45 mm, presenting two distinct thick sections. These regions are prone to hot spots and shrinkage defects due to their slower solidification rates. In a typical sand casting foundry, such design features require careful riser and gate design to ensure directional solidification and adequate feeding. Our initial analysis, without any gating system, revealed two major hot spots: one at the top and another at the lower flange. These locations corresponded to predicted shrinkage porosity, which we aimed to address through process design.
We utilized the ProCAST software for all simulations. The casting material is ZTC4 titanium alloy, with its chemical composition listed in Table 1. The sand mold material was bauxite, preheated to 200°C. The pouring time was set to 6 seconds under static pouring conditions. The thermophysical properties of the alloy, including thermal conductivity, density, enthalpy, and viscosity, were calculated using the Scheil model within the software based on the input composition. These properties are critical for accurate modeling of fluid flow and heat transfer during the casting process in the sand casting foundry.
| Element | Composition (wt.%) |
|---|---|
| Al | 5.5 – 6.8 |
| V | 3.5 – 4.5 |
| Fe | ≤ 0.30 |
| Si | ≤ 0.15 |
| C | ≤ 0.10 |
| N | ≤ 0.05 |
| H | ≤ 0.015 |
| O | ≤ 0.20 |
| Ti | Balance |
To ensure the accuracy and efficiency of our simulations, we performed a mesh independence study. We tested four different mesh sizes: 3 mm, 6 mm, 9 mm, and 12 mm. The temperature evolution at a specific point within the casting was compared, using the 3 mm mesh result as the reference. The simulation time increased exponentially with finer meshes. For example, the 3 mm mesh required over 20 hours, while the 9 mm mesh needed less than 4 hours. The temperature error was minimal for the 6 mm and 9 mm meshes compared to the 3 mm standard. Based on this analysis, we selected a mesh size of 6 mm for the casting body, 12 mm for the pouring cup, and 9 mm for the gates and risers as a balance between accuracy and computational cost. This approach is standard in the sand casting foundry simulation workflow.
| Mesh size (mm) | Computation time (hours) | Temperature error (%) |
|---|---|---|
| 3 | 22.1 | Reference |
| 6 | 7.8 | 0.5 |
| 9 | 3.6 | 1.2 |
| 12 | 1.9 | 3.8 |
In the sand casting foundry, understanding the inherent shrinkage behavior of the bare casting is essential for effective riser design. We analyzed the solidification field of the casting without any gating system. The simulation results indicated two isolated hot spots, designated as Hot Spot A at the top and Hot Spot B at the lower flange. The shrinkage porosity distribution directly correlated with these hot spots. A large shrinkage cavity (Shrinkage A) was observed at the top, and another (Shrinkage B) was found at the lower flange, accompanied by a linear shrinkage zone (Shrinkage C). This analysis guided us in designing two initial process schemes.
Two preliminary gating and riser designs were proposed to feed these hot spots. In Scheme 1, the casting was oriented in its original position. The gating system was designed to feed Hot Spot B directly, while a riser was placed on top to feed Hot Spot A. In Scheme 2, the casting was inverted, placing the thick lower flange section at the top to be fed by a riser. Both designs were evaluated using simulation and subsequent trial production in the sand casting foundry.
The simulation of Scheme 1 showed that when the solid fraction reached 53%, the feeding channel through the gate became blocked. This resulted in a shrinkage cavity near the gate with a volume of 3.9 cm³ and a linear shrinkage defect in the remote area with a volume of 1.0 cm³. The simulation of Scheme 2 indicated that the solidification rate within the hot spot was slower than in the runners. When the solid fraction reached 40%, the feeding channel was already interrupted. This led to a large shrinkage cavity extending 14.5 mm into the casting body, with a total volume of 38 cm³, and a small cavity of 0.4 cm³ in a remote area. Table 3 summarizes the simulation results for both initial schemes.
| Scheme | Shrinkage location | Shrinkage volume (cm³) | Penetration depth (mm) |
|---|---|---|---|
| Scheme 1 | Near gate | 3.9 | N/A |
| Scheme 1 | Remote area | 1.0 | N/A |
| Scheme 2 | Gate root | 38.0 | 14.5 |
| Scheme 2 | Remote area | 0.4 | N/A |
We proceeded with trial production for both schemes in the sand casting foundry. The results from X-ray inspection of the castings from Scheme 1 confirmed the presence of a large shrinkage cavity near the gate and an extensive linear shrinkage defect in the remote area, matching the simulation predictions. The castings from Scheme 2 exhibited a deep shrinkage cavity at the gate root, also consistent with the simulation. The location accuracy of the simulation was high, but the actual shrinkage volume was larger than predicted. We attributed this discrepancy to gas evolution from the sand mold, which was not accounted for in the initial simulation setup.
To improve the predictive accuracy of our numerical model, we calibrated the macroporosity formation parameter (MACROFS) in ProCAST. This parameter dictates the critical solid fraction at which feeding flow ceases, leading to macro-shrinkage. By matching the simulated and experimental results, we adjusted the threshold to reflect the actual foundry conditions more accurately. The improved model was then used to design an optimized gating system, which was based on the Scheme 2 configuration. The key idea was to allow the shrinkage cavity to form entirely within the ingate rather than penetrating into the casting body.
The optimized design featured a conical ingate. Based on the solidification gradient angle observed in the initial simulations, we set the taper angle of the ingate to 25°. Furthermore, based on the depth of shrinkage penetration observed in Scheme 2 (14.5 mm), the height of the ingate was increased to 155 mm to ensure that the final solidification point occurred within the runner. This design promotes directional solidification towards the ingate. The governing heat transfer equation for solidification in our simulation is:
$$ \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \dot{Q} $$
where $\rho$ is density, $C_p$ is specific heat, $T$ is temperature, $t$ is time, $k$ is thermal conductivity, and $\dot{Q}$ is the latent heat source term. Additionally, the criteria for feeding interruption in our model is given by:
$$ f_s > f_{s, crit} $$
where $f_s$ is the solid fraction and $f_{s, crit}$ is the critical solid fraction determined by the calibrated MACROFS parameter.
The simulation results for the optimized design showed that the feeding channel remained open until the casting reached a solid fraction of approximately 76%. This extended feeding time allowed the hot spots to be adequately fed by the molten metal in the ingate. As a result, the final solidification point was located entirely within the ingate. The shrinkage cavity was fully contained within this runner, and the casting body itself was predicted to be free of any significant shrinkage defects. Table 4 compares the key parameters of the initial and optimized designs.
| Parameter | Scheme 2 (Initial) | Optimized Design |
|---|---|---|
| Ingate shape | Rectangular | Conical |
| Ingate taper angle (°) | 0 | 25 |
| Ingate height (mm) | Original | 155 |
| Critical solid fraction for feeding (%) | ~40 | ~76 |
| Shrinkage volume in casting (cm³) | 38.0 | 0 |
We implemented the optimized design in the sand casting foundry for production. The castings underwent X-ray inspection and subsequent machining. The results demonstrated that the casting body was free from major shrinkage cavities. The internal quality met the B-grade standard of GJB 2896A-2020, which is a stringent requirement for aerospace components. No extensive welding repair was needed. This success highlights the effectiveness of our simulation-driven design approach.
In conclusion, our study successfully optimized the sand casting process for an annular titanium alloy casting in a sand casting foundry. The key findings and conclusions are summarized below:
- Numerical simulation accurately predicted the location of shrinkage defects in titanium alloy sand castings, but the predicted volume was less than the actual volume due to mold gas evolution.
- Calibrating the MACROFS parameter based on experimental results significantly improved the accuracy of the simulation model.
- An optimized design featuring a conical ingate with a 25° taper angle and an increased height of 155 mm successfully relocated the shrinkage cavity from the casting body into the ingate.
- The final castings met the stringent B-grade standard of GJB 2896A-2020, eliminating the scrap rate caused by shrinkage porosity.
- The methodology provides a robust forward-design principle for gating systems in the sand casting foundry, using solidification gradient and penetration depth as key parameters.