The pursuit of high-performance and lightweight engineering components, particularly in demanding sectors like aerospace, has led to the widespread adoption of complex investment cast parts. A significant challenge in manufacturing these intricate casting parts is the inherent “variable cross-section effect.” This refers to substantial variations in wall thickness within a single component, which disrupts uniform solidification and heat dissipation. During the solidification of such casting parts, thicker sections act as thermal hubs, cooling at a slower rate than adjacent thinner sections. This differential cooling creates isolated liquid pools within the mushy zone. These pools are often inadequately fed by the remaining liquid metal, leading to the formation of internal shrinkage defects—namely, macro-shrinkage cavities and micro-porosity. Accurately predicting these defects is therefore a cornerstone of developing robust casting processes for high-integrity casting parts.
Our study focuses on the numerical simulation and analysis of these shrinkage phenomena in casting parts with variable cross-sections. We employ a finite element method (FEM) framework to model the investment casting process. A critical aspect of such simulation is the selection of an appropriate porosity prediction criterion. In this investigation, we compare and evaluate several established models. First, the widely used Porosity criterion, which is effective for predicting gross macro-shrinkage. Second, the Niyama criterion and its dimensionless derivative for micro-porosity. Third, a more comprehensive Advanced Porosity Model (APM). The goal is to determine which model offers the highest fidelity when compared against experimental results for a specific alloy system.
The thermal and physical dynamics during solidification of casting parts are governed by the classical heat transfer equation with a source term for the latent heat of fusion:
$$ \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \rho L \frac{\partial f_s}{\partial t} $$
where \( \rho \) is density, \( C_p \) is specific heat, \( T \) is temperature, \( t \) is time, \( k \) is thermal conductivity, \( L \) is latent heat, and \( f_s \) is the solid fraction. The evolution of \( f_s \) with temperature is typically described by a lever rule or Scheil-Gulliver model depending on the alloy’s solidification behavior.
Shrinkage Prediction Criteria: Mathematical Formulations
The formation of porosity in casting parts is a direct consequence of volumetric contraction during the liquid-to-solid phase transition, compounded by insufficient feeding. The various criteria attempt to model this complex interdependency of thermal parameters and fluid flow.
1. The Niyama Criterion and Its Extension
The Niyama criterion is an empirical relationship that correlates the local thermal conditions at the end of solidification with the propensity for micro-porosity formation in casting parts. It is expressed as:
$$ Ny = \frac{G}{\sqrt{R}} $$
where \( G \) is the temperature gradient (\(^\circ\text{C/m}\)) and \( R \) is the cooling rate (\(^\circ\text{C/s}\)). A region is predicted to be prone to shrinkage porosity if \( Ny \) falls below a critical threshold value \( K \), which is alloy-dependent. This criterion essentially identifies regions where a low thermal gradient and a high cooling rate hinder interdendritic feeding.
Carlson and Beckermann extended this concept to predict the actual volume fraction of micro-porosity, \( f_p \). Their dimensionless model is given by:
$$ N_y^* = C_\lambda R^{-1/3} N_y $$
$$ f_p = \beta’ (f_{l,cr}) $$
$$ \text{where } I(f_l, f_{l,cr}) = \frac{f_l}{f_{l,cr}} $$
Here, \( C_\lambda \) is a coarsening constant related to secondary dendrite arm spacing, \( f_{l,cr} \) is a critical liquid fraction for pore nucleation, and \( \beta’ \) is a scaling factor. This model allows for a quantitative estimate of micro-porosity distribution in the final casting parts.
2. The Advanced Porosity Model (APM)
The APM provides a more physics-based approach by coupling the pressure drop in the mushy zone with the nucleation and growth of pores. It is particularly suited for simulating casting parts where gas solubility and pressure conditions vary. The model solves for liquid pressure \( P_l \):
$$ P_l = P_a + P_m + P_d $$
where \( P_a \) is the ambient or applied pressure, \( P_m = \rho_l g h \) is the metallostatic pressure, and \( P_d \) is the dynamic pressure from fluid flow in the mushy zone. The interdendritic flow velocity \( v_l \) is described by Darcy’s law:
$$ g_l v_l = -\frac{K}{\mu} \nabla P_d = -\frac{K}{\mu} [\nabla P_l – \rho_l g] $$
The permeability \( K \) of the dendritic network, crucial for this calculation, is often modeled using the Kozeny-Carman equation:
$$ K(f_s, \lambda_2) = \frac{\lambda_2^2}{180} \frac{(1 – f_s)^3}{f_s^2} $$
where \( \lambda_2 \) is the secondary dendrite arm spacing and \( f_s \) is the solid fraction.
Pore nucleation occurs when the local liquid pressure falls below a critical nucleation pressure \( P_{nuc} \):
$$ P_l \leq P_{nuc} = P_g – \Delta P_r $$
$$ \text{and } \Delta P_r = \frac{2 \sigma_{lg}}{r} $$
Here, \( P_g \) is the total partial pressure of dissolved gases, \( \Delta P_r \) is the Laplace overpressure due to capillarity, \( \sigma_{lg} \) is the liquid-gas surface energy, and \( r \) is the pore curvature radius. For shrinkage-dominated porosity in vacuum casting, \( P_g \) can often be neglected. After nucleation, APM tracks the growth of the pore volume fraction by solving mass conservation equations on a refined sub-grid within the finite element mesh, making it highly detailed for simulating defects in complex casting parts.
Finite Element Modeling Setup for Comparative Analysis
To evaluate the predictive capabilities of the different criteria, we established a simulation model representative of a gravity investment casting process. The model features a casting part with a pronounced variable cross-section, where a thick cylindrical section is connected to a much thinner one. The key parameters for the simulation are summarized below.
| Parameter | Value / Specification | Remarks |
|---|---|---|
| Alloy | Nickel-based Superalloy | Similar to K439B; Properties include temperature-dependent thermal conductivity, specific heat, and viscosity. |
| Pouring Temperature | 1775 K (1502 °C) | – |
| Mold Material | Fused Silica | – |
| Mold Pre-heat Temperature | 1173 K (900 °C) | – |
| Ambient Pressure | Vacuum Conditions | Negligible gas pressure effects for shrinkage. |
| Interfacial Heat Transfer Coefficient (HTC) | Temperature-dependent function | HTC varies from ~500 W/m²·K at high temp to ~100 W/m²·K near solidus. |
| Mesh Size (Casting) | ~2 mm (Tetrahedral elements) | Total elements: ~400,000 for casting and mold system. |
Results: Comparison of Prediction Criteria
The simulation results for the casting part with a cross-sectional area ratio of approximately 6:1 are analyzed. The following table summarizes the key findings from the different porosity prediction models for the characteristic thick section of the casting.
| Prediction Criterion | Predicted Porosity Fraction in Thick Section | Morphology of Predicted Defects | Key Advantages | Noted Limitations |
|---|---|---|---|---|
| Porosity Criterion | ~4.2% | Predicts large, elongated macro-shrinkage cavities concentrated at the top of the thick section. Less detail on micro-porosity distribution. | Simple, robust, and computationally efficient for identifying major shrinkage cavities in casting parts. | Primarily qualitative for macro-shrinkage; does not quantitatively predict micro-porosity volume fraction. |
| Niyama (Ny) & Extended Niyama (Ny*) | Ny* predicted an additional ~0.15% micro-porosity. | Identifies regions at risk of micro-porosity, typically surrounding the macro-shrinkage predicted by the Porosity criterion. Provides a quantitative micro-porosity volume. | Good for assessing feeding difficulty and micro-porosity risk zones. Ny* offers a quantitative estimate. | Empirical; threshold value is alloy-sensitive. May not accurately predict the exact shape and size of macro-pores in complex casting parts. |
| Advanced Porosity Model (APM) | ~3.3% (Total porosity) | Predicts a more realistic defect structure: large-volume pores at the very top, transitioning to a conical region of lower porosity fraction (5%-40%), surrounded by a diffuse region of micro-porosity (<1%). | Physics-based, considers pressure drop and feeding flow. Provides a continuous and detailed distribution of porosity fraction, closely matching experimental defect morphology in casting parts. | Computationally more intensive. Requires accurate input parameters for permeability and nucleation. |
Experimental validation was conducted by producing the casting part under the simulated conditions. Radiographic and metallographic inspection of the sectioned casting revealed the actual defect distribution. The experimental results showed large macro-shrinkage cavities at the top of the thick section, with a concentrated zone of micro-shrinkage directly beneath it, tapering off towards the bottom. The morphology and location predicted by the APM criterion showed significantly better agreement with the experimental observations compared to the combined Porosity+Ny* result. The Porosity criterion correctly located the macro-cavity but overstated its volume and predicted an unrealistic elongated shape, while the Ny* criterion under-predicted the extent of the concentrated shrinkage zone.
Parametric Study: Influence of Cross-Sectional Ratio
Having established confidence in the APM model, we performed a parametric study to investigate the systematic effect of the cross-sectional ratio on shrinkage formation in such casting parts. Four different area ratios were simulated: 2:1, 4:1, 6:1, and 12:1. The thick section diameter was varied while keeping the thin section constant.
The key metrics extracted from the APM simulations are the total porosity volume and the average porosity fraction within the characteristic thick section of the casting part. The trends are clearly non-linear and are quantified below.
| Cross-Sectional Area Ratio (Thick:Thin) | Approx. Thick Section Diameter (mm) | Total Porosity Volume in Thick Section (cm³) | Average Porosity Fraction in Thick Section (%) | Time for Defect Initiation (s after pour) | Time for Final Defect Formation (s after pour) |
|---|---|---|---|---|---|
| 2:1 | 18.5 | 0.44 | 2.31 | 50 | 110 |
| 4:1 | 25.4 | 1.52 | 2.98 | 60 | 180 |
| 6:1 | 31.0 | 2.25 | 3.28 | 65 | 250 |
| 12:1 | 44.0 | 3.65 | 4.56 | 80 | 385 |
The data demonstrates a strong correlation between the severity of the variable cross-section and the severity of shrinkage defects in the resulting casting parts. As the ratio increases from 2:1 to 12:1:
- Porosity Volume and Fraction: Both the total pore volume and the average porosity fraction increase significantly. The porosity fraction more than doubles, indicating that larger thermal hubs are not only harder to feed but also experience more severe localized shrinkage.
- Defect Initiation and Evolution Time: The onset of pore nucleation is delayed, and the total time required for the defect structure to fully form increases dramatically. This is because the thicker section takes much longer to cool down to the solidus temperature and progresses through the vulnerable mushy zone stage over an extended period.
- Morphological Evolution: The shape of the porosity zone also changes. For lower ratios (2:1, 4:1), the central region of the porosity cone consists primarily of porosity fractions below 50%. For the highest ratio (12:1), the core of the defect contains porosity fractions exceeding 50%, representing a large, coherent macro-cavity. Furthermore, the overall aspect ratio (length-to-width) of the conical porosity zone decreases, meaning the defect becomes wider and more bulbous relative to its depth as the section thickens.
This behavior can be rationalized by considering the solidification sequence. The thin section solidifies rapidly, cutting off direct feeding paths to the thick section. The thick section then solidifies from the outside in, creating a growing mushy zone. The pressure drop \( \nabla P \) in this zone, critical in the APM equations, becomes more severe for larger sections due to the longer feeding path length \( L_{path} \) and the decreasing permeability \( K(f_s) \) as solidification proceeds. The relationship can be conceptually simplified as:
$$ \nabla P \propto \frac{\mu \dot{V}}{K(f_s) A_{flow}} L_{path} $$
where \( \dot{V} \) is the volumetric shrinkage rate and \( A_{flow} \) is the available flow area. In larger casting parts with bigger thick sections, \( L_{path} \) increases and \( A_{flow} \) in the later stages of solidification becomes extremely small, leading to a steep pressure drop and severe porosity.
Conclusions and Implications for Process Design
This investigation into the numerical simulation of shrinkage defects in variable cross-section casting parts leads to several key conclusions:
- The Advanced Porosity Model (APM) demonstrates superior predictive accuracy for both macro- and micro-shrinkage defects in nickel-based superalloy investment castings compared to the standard Porosity and Niyama criteria. Its physics-based approach, accounting for interdendritic fluid flow and pressure drop, yields a more realistic representation of the defect’s size, shape, and distribution gradient.
- The cross-sectional ratio is a dominant geometric factor influencing shrinkage severity. The quantitative relationships established—increasing porosity volume/fraction and defect formation time with increasing ratio—provide a valuable database for predicting defect levels in new casting parts based on their geometry.
- The evolution of defect morphology, from a narrower, lower-porosity cone to a wider, macro-cavity-dominated zone, offers insights for targeted process optimization. For instance, casting parts with very high cross-sectional ratios will likely require active feeding aids like chills or padding to modify the local solidification pattern, rather than relying solely on riser design.
The methodology and findings presented here form a solid foundation for the virtual development of robust casting processes. By integrating the APM criterion into the simulation-led design cycle, engineers can proactively identify shrinkage risks in complex casting parts, optimize feeder and gating systems, and significantly reduce the time and cost associated with physical trial-and-error methods, ultimately leading to more reliable and high-performance cast components.