The pursuit of enhanced performance and durability in critical automotive components, such as brake drums for heavy-duty vehicles, drives continuous material development. Gray cast iron has been the traditional material of choice due to its favorable combination of good castability, thermal conductivity, damping capacity, and wear resistance. However, under severe service conditions involving repeated thermal cycling and high mechanical stress, standard grades can be prone to thermal fatigue cracking and premature failure. Alloying modifications offer a pathway to improve key properties like high-temperature strength and thermal fatigue resistance. Yet, altering the chemical composition inevitably changes the fundamental material properties—solidification range, fraction solid evolution, density, and thermal characteristics—which directly influence its behavior during the casting process. These changes can shift the locations of thermal centers, alter feeding paths, and introduce new risks for internal casting defect formation, particularly shrinkage porosity. This article presents an integrated methodology combining computational thermodynamics, finite element simulation, and experimental validation to predict, analyze, and control these casting defects in a modified gray cast iron brake drum.
The primary objective is to move beyond trial-and-error in process design. By establishing a reliable digital twin of the casting process that accounts for the modified alloy’s specific properties, we can virtually probe the filling and solidification sequence, identify potential defect sites, and systematically optimize the gating system and process parameters to ensure sound casting quality. The focus is on mitigating shrinkage porosity, a critical internal casting defect that compromises mechanical integrity.
Computational Determination of Modified Alloy Properties
The foundation of an accurate simulation is a correct material model. The modified gray cast iron (MGCI) in this study was derived from a standard HT250 grade by adjusting alloying elements to enhance performance. Key changes included reductions in Ni and Cu, and intentional additions of Ti and B. The precise compositions are detailed in Table 1.
| Material | C | Si | Mn | P | S | Cr | Ni | Cu | Ti | B | Fe |
|---|---|---|---|---|---|---|---|---|---|---|---|
| HT250 (Base) | 3.42 | 1.86 | 0.72 | 0.050 | 0.080 | 0.30 | 0.064 | 0.277 | 0.028 | 0.001 | Bal. |
| Modified (MGCI) | 3.38 | 1.91 | 0.77 | 0.053 | 0.078 | 0.32 | 0.024 | 0.223 | 0.122 | 0.026 | Bal. |
To capture the solidification physics accurately, thermodynamic and property calculations were performed using JMatPro software. A “Back Diffusion” model was employed for the solid-state diffusion, which is crucial for simulating the non-equilibrium solidification conditions prevalent in casting. The results, compared to the base HT250, revealed significant shifts critical for casting defect analysis.
The liquidus and solidus temperatures of the MGCI were calculated to be 1190 °C and 1072 °C, respectively, representing a decrease of approximately 8 °C for both points compared to the base alloy (1198 °C and 1078 °C). This slight widening and lowering of the freezing range is the first indicator of altered solidification behavior. More importantly, the evolution of the fraction of solid ($f_s$) as a function of temperature ($T$) differs markedly, as conceptually described by the following relationship influenced by alloy composition $C_i$:
$$ f_s(T, C_i) = 1 – \left( \frac{T_L(C_i) – T}{T_L(C_i) – T_S(C_i)} \right)^{-\frac{1}{1-k}} $$
where $T_L$ and $T_S$ are the liquidus and solidus temperatures, and $k$ is the partition coefficient. The modified alloy shows a more gradual increase in $f_s$ over a wider temperature interval within the mushy zone. This is illustrated by a flatter $f_s$ vs. $T$ curve, especially noticeable in the mid-range of solidification (around $f_s$ = 0.7). This implies that the modified alloy remains in a pasty, low-strength state for a longer duration, which can affect interdendritic feeding and exacerbate the risk of shrinkage porosity formation if the feeding pressure is inadequate.
Furthermore, the density ($\rho$) variation from liquid to room temperature is affected. The modified alloy exhibits a marginally lower density overall and a slightly smaller total volume change during cooling. The density change during solidification, a key driver for shrinkage, can be related to the fraction solid:
$$ \rho(T) \approx \frac{1}{\frac{f_s}{\rho_s(T)} + \frac{1-f_s}{\rho_l(T)}} $$
where $\rho_s$ and $\rho_l$ are the temperature-dependent solid and liquid densities. The modified alloy’s curve suggests a different pattern of volumetric contraction, which must be correctly modeled to predict shrinkage accurately. These calculated property datasets (enthalpy, thermal conductivity, viscosity, fraction solid, density) were integrated as temperature-dependent functions to create a custom material model for the MGCI, replacing generic database values to ensure simulation fidelity.
Finite Element Model Setup for Process Simulation
The casting process for the brake drum employed green sand molding with a two-cavity layout. The component, with a major diameter of 335 mm and a mass of 25 kg, was oriented with its bowl-shaped section (containing reinforcing ribs) facing downward. The gating system was designed for bottom-side filling. A three-dimensional model of the entire mold assembly, including the sprue, runners, ingates, the two castings, and feeder heads (risers), was constructed.
For computational efficiency and accuracy, a progressive meshing strategy was adopted. The mesh was refined in regions of interest—the castings, gating system, and risers—with element sizes ranging from 1 to 3 mm. The sand mold itself was coarsely meshed, with element size growing progressively to 10 mm at the outer boundaries. This resulted in a high-quality mesh with approximately 4.87 million tetrahedral volume elements and 834 thousand nodes. The governing equations for fluid flow, heat transfer, and solidification were solved using a finite volume approach within the ProCAST simulation environment. The key physics included:
- Fluid Flow: The Navier-Stokes equations with a free surface (Volume of Fluid method).
$$ \rho \left( \frac{\partial \vec{v}}{\partial t} + \vec{v} \cdot \nabla \vec{v} \right) = -\nabla p + \mu \nabla^2 \vec{v} + \rho \vec{g} + \vec{S}_{mushy} $$
where $\vec{v}$ is velocity, $p$ is pressure, $\mu$ is viscosity, $\vec{g}$ is gravity, and $\vec{S}_{mushy}$ is a momentum sink term in the mushy zone that becomes very large as $f_s$ approaches 1, effectively stopping flow in solidified regions. - Heat Transfer: The transient energy equation, accounting for the latent heat of fusion ($L$) release during phase change.
$$ \rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) – \rho L \frac{\partial f_s}{\partial t} $$
where $c_p$ is specific heat and $k$ is thermal conductivity. - Defect Prediction: The “New-APM” (Anatomical Pore Model) criterion was activated to predict shrinkage porosity. This model estimates the pressure drop in the interdendritic liquid and correlates it to pore formation based on a critical pressure threshold.
The initial process parameters for the baseline simulation were set as: pouring temperature = 1350 °C, pouring time = 30 s. Interface heat transfer coefficients (HTC) between the metal and sand mold were defined, and appropriate boundary conditions for convection and radiation to the ambient environment were applied.
Analysis of Filling and Solidification Behavior
The simulation of the baseline process provided detailed insights into the sequence of events leading to potential casting defect formation.
Filling Pattern: The metal entered the mold cavity through the side ingates at the bowl level. The filling sequence was smooth and sequential from the bottom upwards. No turbulent splashing or cold shuts were observed, indicating a well-designed gating system for achieving quiescent filling. The metal front advanced uniformly, minimizing the risk of oxide film entrainment and gas aspiration—common casting defects related to poor filling.
Solidification Sequence & Thermal Analysis: After filling, the critical solidification phase began. The temperature field evolution revealed that solidification initiated at the thin sections: the top flange (exposed to the cooler sand at the mold top) and the extremities of the bottom ribs. However, the thermal center, or “hot spot,” was clearly identified at the junction of the bowl’s thick section and the reinforcing ribs, adjacent to the ingates. This area, being geometrically thick and thermally connected to the still-hot gating system, remained liquid the longest.
The solidification time contour plot quantitatively confirmed this. While the top flange and thin drum walls solidified between 480-590 s, the last points to solidify (beyond 670 s) were the feeder head and, critically, the hot spot in the brake drum’s bowl near the ingate. This created an isolated liquid pool within the casting itself, disconnected from the feeder head’s liquid metal supply because the connecting ingates had already solidified. This violation of directional solidification (casting -> feeder) is a classic precursor to shrinkage porosity.
The progression of the fraction solid ($f_s$) clearly showed this isolation. As solidification advanced, the mushy zone ($0 < f_s < 1$) retreated towards this final hot spot. When $f_s$ in the feeding paths (the ingates) reached 1, fluid flow for feeding ceased, despite the adjacent casting region still having $f_s < 1$. The pressure in this isolated liquid pocket, $P_{local}$, drops due to continued solidification shrinkage, governed by:
$$ \nabla \cdot \vec{v} \approx -\frac{1}{\rho} \frac{d\rho}{dt} \approx \beta \frac{\partial f_s}{\partial t} $$
where $\beta$ is the solidification shrinkage coefficient. When $P_{local}$ falls below a critical threshold (often related to local gas solubility or pore nucleation pressure), a shrinkage cavity or dispersed porosity forms.
Prediction of Shrinkage Porosity: The New-APM module processed this thermal and feeding data to predict the location and severity of shrinkage. The simulation output highlighted a high-risk zone precisely at the identified hot spot on the brake drum’s inner bowl/rib junction. The predicted casting defect was characterized as a region of interconnected microporosity with an average predicted porosity fraction of approximately 3.5% in the affected elements.
Experimental Validation of Simulated Casting Defects
To validate the accuracy of the computational model and its casting defect prediction, actual castings were produced under the same baseline conditions (1350°C pouring temperature, 30 s pour time). The manufactured brake drums were visually inspected and then subjected to non-destructive testing and sectioning.
The experimental results corroborated the simulation findings. The castings exhibited sound external surfaces with no major filling-related defects. However, upon cutting and inspecting the region corresponding to the simulated hot spot, a significant shrinkage cavity was revealed. The location, size, and morphology of the actual defect aligned remarkably well with the simulated porosity prediction. The defect showed characteristic dendritic surfaces and shiny, rounded interiors indicative of solidification shrinkage under low pressure, not gas porosity. This successful validation confirmed that the finite element model, equipped with the customized MGCI material properties, was a reliable tool for predicting this type of internal casting defect.
Systematic Process Optimization to Eliminate Defects
With a validated model, the next step was to virtually test and optimize the process to eliminate the predicted shrinkage. The optimization targeted two main aspects: the gating system geometry and the key process parameters.
1. Gating System Modification: The root cause was the premature solidification of the ingates, which severed the feeding path from the riser. The solution was to increase the thermal mass of the gating system at strategic points to delay its solidification, effectively extending the feeding time. Modifications included:
- Increasing the cross-sectional area (width) of the ingates nearest to the sprue.
- Enlarging the sprue base (runner well) to act as a supplementary thermal reservoir.
These changes were modeled, and the simulation was re-run. The results showed a dramatic improvement. The solidification sequence was altered; the modified gating system now remained liquid longer than the critical hot spot in the casting. Consequently, the last point to solidify shifted from within the casting to the gating system itself, which is acceptable as it is part of the scrap. The predicted shrinkage porosity in the brake drum was virtually eliminated, with only minimal, dispersed porosity predicted (average <1% in few locations). The major casting defect risk was successfully transferred out of the final part.
2. Process Parameter Optimization: Alongside geometrical changes, pouring temperature ($T_{pour}$) and pouring time ($t_{pour}$) are critical variables affecting fluidity, thermal gradients, and feeding efficiency. A design of experiments was conducted using the validated model to minimize the total shrinkage pore volume ($V_{shrink}$). The objective function can be conceptually framed as:
$$ \min_{T_{pour}, t_{pour}} V_{shrink}(T_{pour}, t_{pour}) $$
subject to process constraints (e.g., maximum temperature to avoid sand burn-on, minimum time to avoid mistruns).
The effects are summarized in Table 2 and can be interpreted as follows:
| Pouring Temp. Range | Pouring Time | Effect on Shrinkage & Mechanism | Optimality |
|---|---|---|---|
| Low (1350-1375°C) | All Times | High porosity (>2%). Low superheat leads to high viscosity, short fluid life, and early blockage of feeding channels. | Poor |
| Medium (1375-1400°C) | Short-Medium (25-30s) | Moderate porosity (~1.3%). Improved fluidity and feeding, but thermal mass of gating may still be limiting. | Acceptable |
| Medium (1375-1400°C) | Long (35s) | Higher porosity (>2%). Slow fill leads to excessive heat loss in early-poured metal, reducing effective feeding range. | Poor |
| High (1425°C) | Short-Medium (25-30s) | Lowest porosity (~0.86%). High superheat maintains strong thermal gradient and extends feeding pressure from risers/gating. Optimal fill rate avoids excessive cooling. | Optimal |
| High (1425°C) | Long (35s) | High porosity. Similar detrimental effect as slow pour at medium temperature. | Poor |
The interaction is clear: a higher pouring temperature is beneficial only when coupled with a sufficiently fast pour to deliver that heat effectively to the mold cavity. The optimal combination was found to be $T_{pour} = 1425°C$ and $t_{pour} = 30 s$. This combination, simulated with the modified gating design, resulted in the most favorable thermal conditions for directional solidification and effective interdendritic feeding, thereby minimizing the casting defect potential.
Conclusion and Industrial Verification
This study demonstrates a robust framework for addressing casting defect challenges in advanced alloy development. By first calculating the precise thermophysical properties of a modified gray cast iron, then incorporating them into a detailed finite element model of the casting process, it was possible to accurately predict the formation of shrinkage porosity in a complex brake drum geometry. The model identified the failure of directional solidification due to an undersized gating system as the root cause.
Through virtual optimization, both the gating geometry and process parameters (1425°C pour temperature, 30 s pour time) were improved to ensure the thermal center resided in the feeder/gating system, not the casting. The final optimized process was implemented in production. The resulting castings were fully dense, free from the internal shrinkage casting defect that plagued the initial trials, and exhibited excellent surface quality after machining. This integrated computational materials engineering (ICME) approach significantly reduces development time, material waste, and cost while ensuring the reliable production of high-performance cast components with complex microstructural requirements.