Slag inclusions represent a pervasive and detrimental defect class in cast metal components, arising primarily from incomplete slag removal, reoxidation, or the entrainment of oxide films during mold filling. Analogous to shrinkage pores, these non-metallic discontinuities severely compromise the structural integrity of the metal matrix by acting as potent stress concentrators. In service, slag inclusions often serve as initiation sites for fatigue cracks, significantly reducing the component’s lifespan and reliability. The morphological characteristics—size, distribution, and volume fraction—of these slag inclusions, coupled with the local geometry of the casting, critically influence the fatigue crack initiation and propagation behavior. Therefore, accurate prediction and effective control of slag inclusion formation and distribution are paramount for enhancing casting quality and performance.
During mold filling and solidification, entrained slag particles are transported by the molten metal. Throughout this journey, these particles are not merely passive passengers; they undergo complex interactions including collision, adsorption, and coalescence. This dynamic aggregation leads to particle growth, fundamentally altering their physical characteristics. The coalesced slag clusters, due to their non-wetting nature with the melt, possess lower surface energy and resist fragmentation. Furthermore, their increased volume and mass significantly affect the hydrodynamic forces—drag, buoyancy, and lift—acting upon them, thereby altering their trajectory and final resting position within the solidified casting. Predicting this stochastic process is highly complex, as it is intricately coupled with the transient temperature, velocity, and solidification fields of the molten metal, all influenced by material properties, pouring temperature, mold characteristics, and filling velocity.
Traditional dispersed phase particle tracking models often treat inclusions as inert, non-interacting entities with constant size. This simplification limits prediction accuracy, as it ignores the dynamic evolution of slag inclusion size and its consequent feedback on particle motion. To bridge this gap, this work develops a Particle Aggregation Growth Model (PAGM). The PAGM enhances classic particle tracking by incorporating physics-based rules for the dynamic coalescence and growth of slag particles, providing a more realistic simulation of slag inclusion behavior during casting.
Theoretical Framework of the PAGM
The PAGM is built upon a Eulerian-Lagrangian framework where the molten metal (continuous phase) is governed by the standard conservation equations for mass, momentum, and energy. The slag inclusions (dispersed phase) are tracked as discrete particles whose motion is governed by Newton’s second law, considering forces from the surrounding fluid.
Fundamental Assumptions
To render the complex problem tractable while retaining critical physics, the following assumptions are made within the PAGM:
- All slag inclusions are modeled as solid spheres.
- Upon aggregation, the resulting cluster is also modeled as a solid sphere whose volume is the sum of the volumes of the constituent particles. Once formed, clusters do not break apart.
- Aggregation follows a “larger particle consumes smaller particle” algorithm within an interaction zone. The mass and volume of the smaller particle are added to the larger one, and the smaller particle is removed from the simulation.
- In regions of high fluid shear (flow velocity exceeding a critical threshold), aggregation is suppressed to account for the disruptive forces that can prevent coalescence or break apart loosely bound clusters.
- Momentum is conserved during the inelastic collision and coalescence event, with kinetic energy loss neglected.
Particle Motion and Aggregation Governing Equations
The motion of a single slag particle is determined by the balance of forces, including pressure gradient force, drag force, gravity, and buoyancy. The equation of motion in the k-direction (where k represents x, y, or z) is given by:
$$ \frac{\pi d_p^3 \rho_p}{6} \frac{du_{p,k}}{dt} = (F_{p,k} + F_{g,k}) – (F_{d,k} + F_{b,k}) $$
Where:
- $d_p$ is the particle diameter.
- $\rho_p$ is the particle density.
- $u_{p,k}$ is the particle velocity component.
- $F_{p,k}$ is the force due to the fluid pressure gradient.
- $F_{g,k}$ is the gravitational force component.
- $F_{d,k}$ is the fluid drag force component, typically modeled as $ F_{d,k} = \frac{1}{2} C_D \rho_f A_p |\vec{u_f} – \vec{u_p}| (u_{f,k} – u_{p,k}) $, where $C_D$ is the drag coefficient, $\rho_f$ is the fluid density, $A_p$ is the particle’s cross-sectional area, and $\vec{u_f}$ is the local fluid velocity.
- $F_{b,k}$ is the buoyancy force component.
The aggregation logic is activated during each computational time step only when a pair of particles satisfies three concurrent criteria:
1. Proximity Criterion: The center-to-center distance $L$ between two particles must be within a defined interaction zone, which scales with particle diameter:
$$ L \leq \frac{1}{2} L_{i\_lim} = \frac{1}{2} k d_i $$
Here, $d_i$ is the diameter of the $i^{th}$ particle, and $k$ is an interaction distance coefficient.
2. Absolute Velocity Criterion: To enforce the high-shear suppression rule, a particle’s speed must be below a threshold to be eligible for aggregation:
$$ |\vec{u_{p,i}}| \leq u_{lim} $$
3. Relative Velocity Criterion: The relative motion must indicate a converging trajectory. This is satisfied if either condition holds (shown for the x-direction):
- Converging Motion: Particles moving towards each other.
$$ u_{p,x\_i} \cdot u_{p,x\_j} < 0 \quad \text{and} \quad (x_i – x_j) \cdot (u_{p,x\_i} – u_{p,x\_j}) < 0 $$ - Catching-Up Motion: Particles moving in the same direction, with the faster one approaching the slower one from behind.
$$ u_{p,x\_i} \cdot u_{p,x\_j} > 0, \quad |u_{p,x\_i}| > |u_{p,x\_j}|, \quad \text{and} \quad u_{p,x\_i} \cdot (x_i – x_j) < 0 $$
Analogous conditions are checked for the y and z directions.
The model was implemented via user-defined functions and subroutines developed in Fortran and integrated with a commercial finite-volume based casting simulation software capable of fluid flow, heat transfer, and solidification modeling.
Simulation of Slag Inclusion Behavior in a Propeller Casting
Case Study: ZL102 Aluminum Alloy Propeller
A small marine propeller casting was selected as the simulation and validation case. Key parameters are summarized below.
| Parameter | Value / Description |
|---|---|
| Casting Material | ZL102 Aluminum Alloy |
| Propeller Diameter | 140 mm |
| Mold Type | Sand Mold, Gravity Casting |
| Pouring Temperature | 680°C |
| Pouring Rate | 0.3 kg/s |
| Simulated Slag Particle | SiO₂ (Density: 2.3 g/cm³) |
| Initial Particle Diameter | 0.6 mm (average) |
| Total Particle Mass Introduced | 1.25 g (~4800 particles) |
The simulation tracked the temperature field, fluid flow, and the trajectory and aggregation state of each discrete slag particle from the start of pouring through complete solidification.
Analysis of Slag Inclusion Distribution and Formation Mechanism
The simulation results revealed a clear and physically consistent narrative of slag inclusion transport and capture. During initial filling, the high-velocity region in the sprue prevented significant aggregation due to high shear forces. As the melt entered the wider hub region and diverged into the three blades, the velocity dropped markedly, and flow recirculation zones developed. In these calmer regions, the proximity, velocity, and trajectory criteria were frequently met, initiating the aggregation process among the entrained slag inclusions.
Particles carried along the advancing melt front were often trapped at the thin edges of the blades, particularly near the blade tips, where geometric constraints impeded their motion. After the blades were filled and the melt converged again in the hub, a region of low velocity and high particle concentration formed near the blade roots, creating an ideal environment for extensive collision and coalescence of slag inclusions. The larger aggregated clusters experienced significantly greater buoyancy forces, accelerating their upward movement towards the casting’s top surfaces. Meanwhile, smaller, non-aggregated or slowly growing particles remained suspended within the bulk liquid and were eventually encapsulated by the advancing solidification front, leading to internal slag inclusions.
The final predicted distribution showed a strong tendency for slag inclusions to be located on or near the upper surfaces of the blades and the central hub, with larger clusters found in areas where flow stagnation and low velocity prevailed. This pattern underscores the critical role of fluid dynamics in determining the final location and size of these defects.
Comparative Study: Impact of Gating System Design
The gating system is the first line of defense against slag inclusion defects. To quantitatively assess its influence, four distinct gating system designs were simulated and compared using the PAGM framework: Top-Gating, Bottom-Gating, Bottom-Gating with a Strainer/Creaming Basin, and Bottom-Gating with a Central Riser.
| Gating System Design | Max. Particle Diameter (mm) | Total Final Particle Count* | Particle Count in Casting | Total Slag Volume in Casting (cm³) | Slag Trapping Efficiency (%)** |
|---|---|---|---|---|---|
| Top-Gating | 2.73 | 998 | 299 | 0.131 | 74.74 |
| Bottom-Gating | 2.30 | 1022 | 358 | 0.309 | 42.78 |
| Bottom-Gating with Strainer | 1.89 | 1246 | 236 | 0.198 | 63.33 |
| Bottom-Gating with Central Riser | 2.27 | 1075 | 270 | 0.213 | 60.56 |
| * Includes particles in gating channels and risers. ** Efficiency = (1 – (Slag Volume in Casting / Total Introduced Slag Volume)) × 100% |
|||||
Discussion of Gating System Performance
The analysis yields clear insights into how gating design affects slag inclusion distribution:
1. Top-Gating System: This design resulted in the lowest number and volume of slag inclusions within the actual propeller casting. The sprue itself acted as an effective slag trap. The calm molten metal pool forming at its base allowed for extensive aggregation (hence the largest maximum particle diameter) and retention of slag inclusions. Consequently, it demonstrated the highest slag trapping efficiency (~75%).
2. Bottom-Gating System: While promoting tranquil mold filling, the basic bottom-gating design offered little obstruction to slag particles entering the mold cavity. Most introduced slag inclusions were carried into the casting, leading to the highest volume of internal slag defects and the lowest trapping efficiency (~43%). The simulation confirmed visual observations of significant slag accumulation on the upper surfaces of the blades.
3. Bottom-Gating with a Strainer/Basin: The addition of a centrifugal separation chamber in the runner significantly improved performance. This feature effectively intercepted a substantial portion of the incoming slag inclusions, preventing them from entering the mold cavity. The slag trapping efficiency improved to over 63%. The maximum particle size was smaller than in the top-gate case, suggesting different aggregation dynamics within the basin.
4. Bottom-Gating with a Central Riser: The central riser primarily functions as a feeder but also serves as a slag collection point for particles that enter the casting and subsequently float upwards. It provided a measurable improvement over the plain bottom-gate, reducing the slag volume in the casting and achieving ~61% efficiency. However, it is less effective than a dedicated upstream slag trap (the strainer) because it must collect slag after it has already entered the casting cavity.
The governing principle for the buoyant motion of slag inclusions can be summarized by a balance between drag and net buoyancy during the floating period. The terminal velocity $v_t$ of a spherical inclusion can be approximated by:
$$ v_t = \sqrt{\frac{4d_p g (\rho_f – \rho_p)}{3 C_D \rho_f}} $$
where $g$ is gravitational acceleration. This relationship highlights that larger aggregated particles ($d_p$↑) float faster ($v_t$↑), making them more likely to reach the top surface or riser before being trapped by solidifying metal. The gating system directly influences the opportunity for this aggregation and the availability of safe harbors (like risers or quiet zones) to collect the floating slag inclusions.
Conclusions
This investigation, employing the advanced Particle Aggregation Growth Model (PAGM), provides a comprehensive analysis of slag inclusion behavior in aluminum alloy castings and the critical influence of gating system design. The key findings are:
- Dynamic Growth is Crucial: The process of collision and coalescence among slag inclusions during mold filling is a fundamental phenomenon that continuously alters their size and, consequently, their buoyancy and motion trajectory. Ignoring this dynamic growth, as standard particle models do, limits the predictive accuracy for final slag inclusion distribution.
- Gating System is a Primary Control Lever: The design of the gating system profoundly impacts the severity and location of slag inclusion defects. Top-gating systems can act as effective slag traps for certain geometries, while bottom-gating systems require integrated features like strainers, filters, or well-placed risers to achieve satisfactory slag exclusion.
- Quantitative Predictive Capability: The PAGM framework offers a significantly enhanced tool for simulating slag inclusion formation. By accounting for aggregation, it provides more reliable predictions of not only the location but also the characteristic size of slag inclusions, moving beyond simple binary defect prediction.
- Design Optimization Guidance: The model enables a quantitative comparison of different gating designs in terms of slag trapping efficiency before physical prototyping. This capability is invaluable for optimizing casting processes to minimize slag-related defects, improve yield, and enhance the mechanical reliability of cast components.
In summary, the integration of a physics-based particle aggregation model into casting simulation represents a substantial step forward in defect prediction technology. It offers foundry engineers a powerful virtual tool to design gating and feeding systems that actively control and minimize the detrimental effects of slag inclusions, leading to the production of higher-integrity castings.