The transformation of the casting industry is intrinsically linked to advanced computer-aided engineering. The computer simulation of the casting process is now an internationally recognized frontier in manufacturing and materials science and represents the essential path for modernizing traditional foundry practices. The inherent complexity of solidification often leads to the formation of detrimental casting defects such as shrinkage cavities, shrinkage porosity, and hot tears. These casting defects pose a significant threat to the structural integrity, mechanical performance, and service life of the final component. Consequently, the ability to predict these imperfections before physical production is paramount. Numerical simulation provides a powerful tool for this purpose, enabling the virtual analysis of filling, solidification, and stress development to forecast potential casting defect locations. This proactive approach allows for the optimization of product design and manufacturing parameters, serving as a critical pathway to ensuring casting quality. In our research, we have focused on the coupled numerical simulation of the transient temperature and stress fields during solidification to predict the probable locations of shrinkage-related casting defects and assess the risk of hot tearing.
Theoretical Foundation: Governing Equations for Heat Transfer
The accurate prediction of a casting defect like shrinkage or hot tearing begins with a precise calculation of the transient temperature field. In a three-dimensional domain, the temperature field variable \( T(x, y, z, t) \) must satisfy Fourier’s law of heat conduction combined with the principle of energy conservation. The governing differential equation in Cartesian coordinates is given by:
$$
c\rho \frac{\partial T}{\partial t} – k \left( \frac{\partial^2 T}{\partial x^2} + \frac{\partial^2 T}{\partial y^2} + \frac{\partial^2 T}{\partial z^2} \right) = g \quad \text{(within the domain } \Omega \text{)}
$$
Where:
\( T \) is the temperature (°C),
\( t \) is time (s),
\( c \) is the specific heat capacity (J/(kg·°C)),
\( \rho \) is the density (kg/m³),
\( k \) is the thermal conductivity (W/(m·°C)),
\( g \) is the internal heat generation rate (W/m³).
This equation represents a balance: the rate of internal energy increase (left term) equals the net heat conduction into the differential volume plus any internal heat generation. The solution of this transient problem requires appropriate boundary and initial conditions (\( T = T_0 \) at \( t = 0 \)). The boundary conditions are classified into three types:
1. Dirichlet (First-type) Boundary Condition:
Specifies the temperature on the boundary \( \Gamma \).
$$
T|_{\Gamma} = f(x, y, z, t)
$$
2. Neumann (Second-type) Boundary Condition:
Specifies the heat flux across the boundary. Here, \( q \) is the imposed heat flux (W/m²).
$$
k \frac{\partial T}{\partial n}|_{\Gamma} = q
$$
3. Robin (Third-type) Boundary Condition:
Specifies convection (and optionally radiation) to an environment at temperature \( T_{\infty} \), with \( h \) as the heat transfer coefficient (W/(m²·°C)).
$$
k \frac{\partial T}{\partial n}|_{\Gamma} = h (T – T_{\infty})
$$
Numerical Methodology: Finite Element Implementation
The simulation of casting solidification involves highly nonlinear transient thermal analysis, as material properties are strongly temperature-dependent. Two critical aspects in this numerical modeling are the handling of latent heat release during phase change and the definition of realistic boundary conditions.
Latent Heat Treatment: The release of latent heat \( L \) (J/kg) during the liquid-to-solid transformation significantly affects the cooling curve. In our finite element model, this is accounted for via the enthalpy method. Enthalpy \( H \) (J/m³), a function of temperature, inherently incorporates both specific heat and latent heat effects. The relationship is defined as:
$$
H(T) = \int_{T_{ref}}^{T} \rho c(T) \, dT + f_s L \rho
$$
where \( f_s \) is the solid fraction. The ANSYS software environment allows for the direct input of temperature-dependent enthalpy data, providing a robust way to model the phase change without explicitly tracking the solidification front.
Boundary Conditions and Model Setup: The heat transfer at the mold-air interface involves both convection and radiation. For simplification at lower mold surface temperatures, the radiation effect can be linearized and incorporated into an effective convection coefficient. In our analysis, different heat transfer coefficients were applied to the mold surfaces: top, bottom, and sides.
The finite element model for a gray iron bracket casting and its sand mold was built using thermal solid elements (SOLID70 in ANSYS). The initial conditions were set as:
– Casting: \( T_{initial} = 1500 \, ^{\circ}\text{C} \) (pouring temperature).
– Sand Mold: \( T_{initial} = 30 \, ^{\circ}\text{C} \).
The material properties for both the casting and the mold were defined as nonlinear functions of temperature. A summary of key temperature-dependent properties for the gray iron used in the simulation is presented in the table below.
| Temperature (°C) | Thermal Conductivity, \( k \) (W/m·°C) | Specific Heat, \( c \) (J/kg·°C) | Density, \( \rho \) (kg/m³) | Enthalpy, \( H \) (x10⁹ J/m³) |
|---|---|---|---|---|
| 25 | 48.0 | 420 | 7100 | 0.075 |
| 500 | 40.5 | 670 | 7000 | 1.55 |
| 750 | 35.0 | 750 | 6950 | 2.65 |
| 1150 (Solidus) | 30.0 | 850 | 6900 | 4.20 |
| 1150 (Liquidus) | 30.0 | 850 | 6800 | 4.95 |
| 1350 | 32.0 | 800 | 6750 | 5.90 |
| 1500 | 34.0 | 750 | 6700 | 6.60 |
Simulation Results: Temperature Field and Prediction of Shrinkage Defects
The primary goal of thermal analysis is to predict shrinkage-related casting defects. Shrinkage cavities and porosity form due to inadequate liquid metal feedingsolidification contraction. A widely accepted criterion for predicting the tendency for such casting defects is the Niyama criterion \( G / \sqrt{R} \), where \( G \) is the temperature gradient (°C/m) and \( R \) is the cooling rate (°C/s) at the solidus temperature or a critical fraction solid.
The criterion is expressed as:
$$
\frac{G}{\sqrt{R}} = \frac{G}{\sqrt{(T_2 – T_1) / (t_2 – t_1)}}
$$
where \( T_1 \) and \( T_2 \) are temperatures at a specific node at times \( t_1 \) and \( t_2 \), respectively. A lower value of \( G / \sqrt{R} \) indicates a higher risk of shrinkage porosity. Empirical studies suggest a critical threshold value of approximately 0.7 °C¹/²·s¹/²/m for many alloys. Regions where \( G / \sqrt{R} < 0.7 \) are prone to shrinkage casting defects, while regions where \( G / \sqrt{R} \geq 0.7 \) are considered sound.
From our transient thermal simulation of the bracket casting, we extracted temperature and temperature gradient data at specific nodes for analysis. The following table summarizes the calculation for selected nodes at two different time steps to compute the Niyama criterion.
| Node ID | Temp. at t=44s, \( T_{44} \) (°C) | Gradient \( G \) (°C/m) | Temp. at t=3500s, \( T_{3500} \) (°C) | ΔT (°C) | Δt (s) | Cooling Rate \( R \) (°C/s) | \( G / \sqrt{R} \) (°C¹/²·s¹/²/m) | Prediction |
|---|---|---|---|---|---|---|---|---|
| 1682 | 152.85 | 6.91 | 144.71 | 8.14 | 500 | 0.0163 | 0.53 | Defect Likely |
| 1683 | 198.24 | 8.00 | 184.10 | 14.14 | 500 | 0.0283 | 0.47 | Defect Likely |
| 1684 | 235.71 | 9.61 | 214.89 | 20.82 | 500 | 0.0416 | 0.48 | Defect Likely |
| 2311 | 287.39 | 19.33 | 265.32 | 22.07 | 250 | 0.0883 | 0.64 | Defect Likely |
| 486 | 196.83 | 12.17 | 195.13 | 1.70 | 250 | 0.0068 | 1.52 | Sound |
| 1663 | 328.42 | 98.77 | 320.60 | 7.82 | 250 | 0.0313 | 5.49 | Sound |
The analysis clearly identifies nodes 1682, 1683, 1684, and 2311 as critical locations with \( G / \sqrt{R} \) values significantly below the 0.7 threshold. These nodes, clustered in the thicker sections and junctions of the bracket, are predicted to be high-risk zones for shrinkage porosity or micro-shrinkage. This numerical prediction was validated against actual production castings, and the locations showed excellent correlation with observed shrinkage casting defects.
Thermal Stress Analysis and Hot Tearing Prediction
Another critical class of casting defect is the hot tear or hot crack, which forms during the final stages of solidification when the coherent solid skeleton has low strength and ductility. If thermal contraction is constrained (by the mold, core, or cooler sections of the casting itself), tensile stresses develop. When these stresses exceed the high-temperature strength of the material, a hot tear initiates. Predicting this casting defect requires a coupled thermo-mechanical simulation to determine the stress/strain fields during solidification.
The total strain \( \varepsilon_{total} \) can be decomposed into elastic, plastic, and thermal components:
$$
\varepsilon_{total} = \varepsilon_{elastic} + \varepsilon_{plastic} + \varepsilon_{thermal}
$$
Where the thermal strain is given by \( \varepsilon_{thermal} = \alpha (T – T_{ref}) \), with \( \alpha \) being the coefficient of thermal expansion. The stress \( \sigma \) is related to the elastic strain by Hooke’s law, \( \sigma = D \varepsilon_{elastic} \), where \( D \) is the elasticity matrix which is itself highly temperature-dependent.
Several criteria have been proposed to predict hot tearing. A practical and operable criterion is based on the von Mises equivalent stress \( \sigma_{vm} \) and the high-temperature tensile strength \( \sigma_b(T) \) of the alloy. The criterion can be expressed as a stress-to-strength ratio \( K_w \):
$$
K_w = \frac{\sigma_{vm}}{\sigma_b(T)}
$$
A hot tear is likely to occur if \( K_w \geq 1 \) in the vulnerable mushy zone temperature range. The major challenge is defining \( \sigma_b(T) \), which drops dramatically near the solidus temperature.
To establish this relationship for the gray iron under study, we first simulated a standard “stress bar” casting of the same material. From the stress analysis at various time steps, we extracted the temperature and corresponding von Mises stress \( \sigma_{vm} \) at numerous nodes. By fitting the data, we derived an empirical linear relationship for the high-temperature tensile strength as a function of temperature for this specific alloy:
$$
\sigma_b(T) = \frac{1}{16} (1500 – T) \quad \text{[MPa, for T in °C]}
$$
This relationship was then applied to analyze the stress field in the main bracket casting. The mechanical simulation was performed with temperature-dependent Young’s modulus, Poisson’s ratio, and yield strength, using the previously calculated temperature history as the thermal load. The results for critical time steps in the vulnerable temperature range (500°C to 1100°C) are summarized below.
| Time (s) | Node Temperature (°C) | Von Mises Stress, \( \sigma_{vm} \) (MPa) | High-Temp Strength, \( \sigma_b(T) \) (MPa) | Stress-to-Strength Ratio, \( K_w \) | Hot Tear Risk |
|---|---|---|---|---|---|
| 92 | 1106.9 | 0.117 | 24.6 | 0.005 | Very Low |
| 117 | 1075.7 | 0.211 | 25.0 | 0.008 | Very Low |
| 242 | 920.0 | 0.284 | 34.0 | 0.008 | Very Low |
| 442 | 720.0 | 0.548 | 46.0 | 0.012 | Low |
| 542 | 635.0 | 0.610 | 51.0 | 0.012 | Low |
| 642 | 580.0 | 0.726 | 57.5 | 0.013 | Low |
| 742 | 538.0 | 0.900 | 60.6 | 0.015 | Low |
The results indicate that throughout the solidification and cooling process, the maximum developed von Mises stress remains orders of magnitude lower than the corresponding high-temperature tensile strength of the material at every analyzed node and time step. The maximum \( K_w \) value observed was 0.015, far below the critical threshold of 1.0. Therefore, the simulation predicts a very low risk of hot tearing for this specific bracket casting under the modeled conditions, which was consistent with the absence of this casting defect in production.
Discussion and Practical Implications
The successful application of finite element-based numerical simulation for predicting casting defects hinges on several factors. First, the accuracy of the input data—temperature-dependent thermal and mechanical properties, boundary conditions, and latent heat parameters—is critical. The use of the enthalpy method proved effective for handling the nonlinearity introduced by phase change. Second, the choice of prediction criteria is crucial. The Niyama criterion (\( G / \sqrt{R} \)) provided a reliable and quantitatively clear method for locating shrinkage porosity, a common and costly casting defect. For hot tearing, the stress-based criterion using a temperature-dependent strength function offered a physically sound approach, though the definition of \( \sigma_b(T) \) remains alloy-specific and often requires calibration from experiments or specialized tests.
The process demonstrated here forms a closed-loop framework for casting quality assurance: Design -> Numerical Simulation -> Defect Prediction -> Design/Process Modification -> Re-simulation. By identifying potential casting defect locations virtually, engineers can proactively implement corrective measures such as:
- Relocating or redesigning feeders and risers to improve feeding for shrinkage-prone areas.
- Modifying chills or cooling channels to alter local solidification patterns and increase \( G / \sqrt{R} \).
- Adjusting gating system design to minimize thermal gradients that induce high stress.
- Altering alloy composition or pouring temperature to change solidification characteristics.
This virtual trial-and-error process drastically reduces the need for expensive and time-consuming physical prototyping, accelerates time-to-market, and significantly improves first-pass yield rates by mitigating casting defects before production begins.
Conclusion
Computer simulation of the casting solidification process, grounded in the finite element method, has matured into an indispensable tool for the modern foundry. This study demonstrated a comprehensive methodology for predicting major casting defects. By performing a nonlinear transient thermal analysis, we obtained the temperature and temperature gradient fields, which enabled the accurate prediction of shrinkage cavity and porosity locations using the Niyama criterion. Subsequently, a coupled thermo-mechanical stress analysis was conducted. An empirical model for the high-temperature tensile strength of the alloy was derived and used in conjunction with the calculated stress fields to assess the risk of hot tearing via a stress-to-strength ratio. The simulation predictions for shrinkage showed strong agreement with actual casting defect locations found in production, while the analysis confirmed a low hot tearing propensity for the studied component. The integrated approach outlined here provides a powerful and practical framework for engineers to visualize solidification, forecast potential casting defects, and iteratively optimize designs and processes. This represents a fundamental shift from reactive defect correction to proactive casting defect prevention, ultimately ensuring higher quality, reliability, and cost-effectiveness in cast component manufacturing.