The research presented in this dissertation focuses on the numerical simulation and process optimization of the lost foam casting process for ductile iron lift arms, which serve as critical lifting components in agricultural tractor hydraulic lift systems. The motivation for this research stems from the recognized capability of casting simulation technology to reduce development costs and enhance quality control, while acknowledging that successful applications to complex three-dimensional castings using the lost foam casting process remain relatively scarce. To address this technological gap, I conducted comprehensive research employing UG software for three-dimensional modeling and ProCAST alongside Visual Environment software for numerical simulation, specifically targeting the challenges of simulating lost foam castings with complex geometries, pouring system design, and riser placement.
Lost foam castings, also known as evaporative pattern casting (EPC), represent a near-net-shape manufacturing technology characterized by the use of a foam pattern that vaporizes upon contact with molten metal. The fundamental process involves forming a foam pattern cluster, applying a refractory coating, embedding it in unbonded dry sand, compacting through three-dimensional vibration, and subsequently pouring molten metal under vacuum conditions. The liquid metal progressively decomposes the foam pattern and occupies the cavity, eventually solidifying into a casting that replicates the pattern geometry with high precision. This technique offers notable advantages including superior dimensional accuracy, surface quality, design flexibility, and the capacity to integrate multiple components into a single casting. Consequently, lost foam castings have found widespread applications in the automotive industry, pipeline components, and complex enclosures, utilizing a broad spectrum of metallic materials, with ductile iron representing one of the most significant material categories for structural applications.
Ductile iron, particularly the QT600-3 grade investigated in this study, offers an exceptional combination of strength and toughness. The QT600-3 designation corresponds to a pearlitic ductile iron with a minimum tensile strength of 600 MPa, yield strength of 370 MPa, and elongation of 3%. The material’s chemical composition, summarized, includes carbon content around 3.7%, silicon at 2.2%, manganese at 0.45%, and trace alloying elements such as chromium, nickel, and copper. The use of lost foam castings for producing ductile iron components presents specific challenges, notably the tendency for carbon-related surface defects and internal shrinkage porosity. The complex thermal and physical interactions during mold filling and solidification require sophisticated simulation tools and proper parameter calibration to achieve reliable predictions of defect formation.
The structure of this dissertation is organized as follows. The research initially explores the special parameter settings required for simulating lost foam casting processes and shrinkage porosity formation in ductile iron castings. This includes analyzing how different parameter configurations affect simulation accuracy and establishing a general methodology for employing ProCAST in the simulation of lost foam castings. Subsequently, the study conducts an in-depth analysis of two trial pouring schemes for the ductile iron lift arms: bottom gating and top gating systems. Through careful comparison of simulated results with actual trial production outcomes, I aim to establish a precise thermophysical model that accurately represents the complex thermal and physical phenomena occurring during filling and solidification in lost foam castings. Based on these findings, I develop an improved gating system design incorporating a step gating approach, which is then further optimized by introducing pre-buried chromite sand cores in critical locations. The ultimate objective is to minimize internal shrinkage defects and achieve high-quality lost foam castings suitable for mass production.
Numerical Simulation Theory of Lost Foam Casting
The numerical simulation of lost foam castings requires a solid understanding of the underlying physical principles governing both mold filling and solidification processes. The filling characteristics of lost foam castings differ substantially from those of conventional cavity casting methods. The presence of foam pattern creates a complex scenario where the molten metal front must continuously decompose the foam, generating gaseous and liquid degradation products that must escape through the refractory coating and dry sand. This process is accompanied by heat transfer, fluid flow, chemical reactions, and potential premature solidification at the flow front.
During the filling of lost foam castings, the molten metal advances in a radial pattern from the ingate rather than being primarily governed by gravitational forces. The flow is controlled by the foam degradation rate, which in turn depends on the thermal content of the molten metal, the density of the foam pattern, coating permeability, vacuum pressure, and the thermal conductivity of the sand mold. The gas gap that forms between the advancing molten metal front and the decomposing foam creates a back pressure that impedes further flow. This pressure must be accurately modeled to predict filling times and identify potential defects such as cold shuts and misruns.
Several numerical models form the foundation for simulating lost foam castings filling. The first fundamental equation is the continuity equation for incompressible Newtonian fluids, which expresses mass conservation within the control volume:
$$\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y} + \frac{\partial w}{\partial z} = 0 \quad (2-1)$$
where $u$, $v$, and $w$ represent the instantaneous velocity components in the x, y, and z directions respectively, and $\rho$ denotes the density of the molten metal in kg/m³. The momentum conservation is expressed through the Navier-Stokes equations, which for incompressible flow with constant viscosity take the form:
$$\frac{\partial u}{\partial t} + u\frac{\partial u}{\partial x} + v\frac{\partial u}{\partial y} + w\frac{\partial u}{\partial z} = F_x – \frac{1}{\rho}\frac{\partial P}{\partial x} + \nu \nabla^2 u \quad (2-2)$$
$$\frac{\partial v}{\partial t} + u\frac{\partial v}{\partial x} + v\frac{\partial v}{\partial y} + w\frac{\partial v}{\partial z} = F_y – \frac{1}{\rho}\frac{\partial P}{\partial y} + \nu \nabla^2 v \quad (2-3)$$
$$\frac{\partial w}{\partial t} + u\frac{\partial w}{\partial x} + v\frac{\partial w}{\partial y} + w\frac{\partial w}{\partial z} = F_z – \frac{1}{\rho}\frac{\partial P}{\partial z} + \nu \nabla^2 w \quad (2-4)$$
where $F_x$, $F_y$, and $F_z$ are the body force components per unit mass, $P$ is the instantaneous pressure, $\nu$ represents the kinematic viscosity, and $\nabla^2$ is the Laplacian operator. Energy conservation during filling must account for the complex heat transfer phenomena at the metal-foam interface, described by the energy equation:
$$\rho C_p \frac{\partial T}{\partial t} = \lambda \nabla^2 T + \rho L \frac{\partial f_s}{\partial t} \quad (2-5)$$
where $T$ is temperature, $C_p$ is specific heat capacity, $\lambda$ is thermal conductivity, $L$ is the latent heat of fusion, and $f_s$ represents the solid fraction.
In lost foam castings, the evolution of gaseous products from foam decomposition creates a localized pressure field that opposes filling. The gas gap pressure model considers the mass balance of gases generated from foam pyrolysis, gases escaping through the coating, and the accumulation of gases within the gap. The pressure at the gap can be expressed as:
$$P_{i+1} = P_0 + \frac{\alpha_P \Delta t (T_m – T_i)}{L_P} \cdot \frac{V_T}{S} – \frac{K F \Delta t (P_i – P_0)}{C_x \delta_i S T_i} + \frac{P_{i+1} \delta_{i+1} – P_i \delta_i}{T_i} \quad (2-6)$$
where $\alpha_P$ is the heat transfer coefficient between the foam and molten metal, $\Delta t$ is the time step, $T_m$ is the melting temperature, $L_P$ is the latent heat of foam degradation, $K$ is the coating permeability, $F$ is the foam perimeter, $S$ is the cross-sectional area, $\delta$ is the gas gap thickness, and $C_x$ is the coating thickness.
The gas gap thickness itself can be estimated from the foam density and filling velocity using:
$$\delta_{i+1} = \delta_i + u\Delta t – \frac{\alpha_P \Delta t (T_i – T_m)}{\rho_P L_P} \quad (2-7)$$
where $u$ is the filling velocity and $\rho_P$ is the foam density. To track the advancement of the free surface during filling, the volume-of-fluid (VOF) method is employed, which introduces the volume fraction function $F$ that satisfies the transport equation:
$$\frac{\partial F}{\partial t} + u\frac{\partial F}{\partial x} + v\frac{\partial F}{\partial y} + w\frac{\partial F}{\partial z} = 0 \quad (2-8)$$
where $F$ ranges from 0 for empty cells to 1 for completely filled cells. The numerical solution of filling in lost foam castings primarily employs the SOLA-VOF method, which couples the SOLA algorithm for pressure-velocity iteration with the VOF method for free surface tracking. This approach has been widely adopted due to its computational efficiency and reasonable accuracy for complex geometires.
Regarding the solidification process simulation for lost foam castings, understanding the unique cooling characteristics is essential. Lost foam castings generally exhibit slower cooling rates compared to conventional sand casting because the dry sand contains no moisture, the refractory coating provides thermal insulation, and the gas gap created during filling further isolates the casting from the mold. This slower solidification affects the overall solidification sequence and heat distribution, potentially promoting coarse grain structures and increased susceptibility to shrinkage defects. The Fourier heat conduction equation forms the basis for solidification modeling, incorporating internal heat generation from latent heat release:
$$\rho C_p \frac{\partial T}{\partial t} = \lambda \left( \frac{\partial^2 T}{\partial x^2} + \frac{\partial^2 T}{\partial y^2} + \frac{\partial^2 T}{\partial z^2} \right) + \rho L \frac{\partial f_s}{\partial t} \quad (2-9)$$
The boundary conditions for heat transfer in lost foam castings include prescribed temperature conditions at the mold surface ($T_w = f(x,y,z,t)$), prescribed heat flux conditions, convective heat transfer at the casting-mold interface expressed as:
$$q = h(T_w – T_f) \quad (2-10)$$
where $h$ is the interface heat transfer coefficient, $T_w$ and $T_f$ represent the temperatures of the mold and metal respectively, and thermal radiation boundary conditions following the Stefan-Boltzmann law:
$$q = \varepsilon \sigma T_s^4 \quad (2-11)$$
where $\varepsilon$ is the emissivity and $\sigma_0$ is the Stefan-Boltzmann constant. The numerical solution methods for the solidification equations include the finite element method (FEM), which offers superior adaptability to complex geometries and precise boundary condition representation. ProCAST utilizes the FEM approach, making it particularly suitable for simulating lost foam castings with intricate three-dimensional geometry.
Simulation Parameter Settings for Lost Foam Castings
The accuracy of numerical simulation for lost foam castings depends critically on proper parameter settings that reflect the unique physical phenomena of the process. Through systematic investigation, I identified the following critical parameters that must be specially configured when using ProCAST for simulating lost foam castings. These parameters are summarized in Table 1 with their recommended values and physical significance. The geometric model for lost foam castings simulation must include at least three components: the empty pouring system, the foam pattern representing the cavity, and the dry sand mold. For the material properties, the pouring system is defined as an empty casting region, the pattern material is defined as foam with appropriate thermal properties including thermal conductivity, density, specific heat, and latent heat of decomposition, and the sand mold is defined as a permeable medium with properties reflecting the unbonded nature of the sand.
The interface between the molten metal and the foam pattern is critically important. In ProCAST, this interface is defined as “EQUIV” to allow simultaneous heat and mass transfer, enabling the thermal interaction between the advancing metal front and the decomposing foam on the lost foam castings simulation. Additional parameters related to foam burning and gas behavior must be specified, including the heat transfer coefficient between the molten metal and foam (FOAMHTC and FOAMHTCMAX), which controls the rate of foam decomposition and thereby the filling progress. These values directly influence the predicted filling time of lost foam castings: higher values lead to faster filling. The parameter BURNZONE defines the characteristic distance for foam burning, which is related to the grid size of the computational mesh.
The gas evolution and pressure effects are modeled through the GAS parameter. For newer versions of ProCAST, setting GAS=2 considers both the existence of gas and its backpressure effect on the molten metal front, which provides a more realistic representation of the filling dynamics in lost foam castings. The parameter PINLET is set to 1 to indicate that pressure boundary conditions control the filling process, reflecting the pressure-driven nature of lost foam filling. The PREF parameter ensures that the differential pressure between the pouring cup and mold exterior accurately converts to the absolute filling pressure, which is essential for simulating the vacuum-assisted filling. Additionally, the refractory coating and adhesive parameters must be defined manually to represent their influence on gas permeability and heat transfer.
| Parameter | Recommended Value | Physical Significance |
|---|---|---|
| FOAMHTC | 0.02 | Heat transfer coefficient between metal and foam |
| FOAMHTCMAX | 0.25 | Maximum heat transfer coefficient at close approach |
| BURNZONE | 1.0 cm | Characteristic distance for foam burning |
| GASFRAC | 0.1 | Fraction of foam converted to gaseous products |
| GAS | 2 | Gas model with backpressure consideration |
| PINLET | 1 | Pressure boundary as filling condition |
| PREF | 1 | Reference pressure adjustment |
For predicting shrinkage porosity in ductile iron lost foam castings, the graphite expansion phenomenon must be considered. ProCAST offers a macroscopic modeling approach coupled with microstructural calculations to capture the density changes associated with graphite precipitation during eutectic solidification. To enable this feature, the porosity calculation parameter POROS must be activated in the thermal module. The key parameters affecting the accuracy of shrinkage porosity prediction in ductile iron include the graphitization degree factor (GRAPHITE), which ranges from 0 to 1, with 1 indicating complete consideration of graphite expansion pressure. A higher GRAPHITE value reduces the predicted shrinkage porosity because graphite expansion partially compensates for liquid contraction.
Another critical parameter is the fading factor (FADING), which accounts for the deterioration of nodularization treatment over time. Ductile iron is typically produced through inoculation with silicon and magnesium agents, and the graphite nodule count and spheroidization efficiency decrease with time after treatment. A smaller FADING value indicates more prolonged and effective graphite expansion, resulting in better self-feeding capacity. For lost foam castings, which exhibit slow cooling, a lower FADING value is recommended to reflect sustained graphite expansion. Additionally, the parameter MGTREAT defines the time from magnesium agent addition to the commencement of solidification, which affects the progress of nodularization fading. The mold rigidity parameter (MOLDRIG) characterizes the stiffness of the mold, taking values from 0 for soft molds to 1 for completely rigid molds. A more rigid mold allows the graphite expansion pressure to be redirected internally, improving the feeding of liquid metal to shrinkage areas.
Finally, the type of sand used in the mold significantly influences the solidification behavior and resulting shrinkage porosity in lost foam castings. Through a comparative analysis, I studied the effect of different sand types: resin sand, silica sand, and chromite sand. The simulation results indicate that chromite sand, with its higher thermal conductivity, promotes faster cooling, reduces porosity volume, and disperses shrinkage porosity. Conversely, silica sand, which is commonly used in lost foam castings, provides slower cooling rates and can lead to more concentrated defects.
Ductile Iron Lift Arms: Casting Process Analysis and Trial Schemes
The engineering component investigated in this research is a lift arm assembly used in agricultural tractor hydraulic systems. These lost foam castings were previously manufactured as three separate parts: an upper arm, a lower arm, and a cylindrical hub connecting them. By exploiting the design flexibility of the lost foam casting process, the component was redesigned as a single integrated casting, as shown in Figure 2. This integration eliminates assembly steps, reduces manufacturing costs, and improves overall performance. The three-dimensional envelope dimensions are 375 mm × 142 mm × 552 mm, and the component weighs approximately 30 kg. The geometry presents significant complexity, featuring curved surfaces, multiple bosses, recesses, transition radii, and an internal stepped cylindrical bore. The wall thickness varies considerably, creating multiple dispersed hot spots, and the middle section of the cylinder wall is relatively thin.
From a casting process standpoint, the lift arms present several challenges. The complex geometry with significant wall thickness transitions creates challenges for directional solidification and feeding. The presence of multiple isolated thermal nodes increases the risk of shrinkage defects. The component serves as a load-bearing structure operating under complex loading conditions, necessitating a dense internal structure free from significant porosity. Surface quality must be impeccable, without sand holes, slag inclusions, or surface folds. The cast material is QT600-3, a pearlitic ductile iron specified by the chemical composition shown in Table 2.
| Element | C | Si | Mn | P | S | Cr | Ni | Cu | Mg |
|---|---|---|---|---|---|---|---|---|---|
| Composition (%) | 3.7 | 2.2 | 0.45 | 0.055 | 0.02 | 0.25 | 0.15 | 0.45 | 0.055 |
The mass production layout was designed as one heat containing six castings in a box, arranged as two castings per gating group, with three groups per box. The castings were positioned vertically to facilitate sand filling into the internal stepped holes and recesses, preventing mold collapse and core movement. The gating systems were designed based on empirical approaches derived from long-term production experience, incorporating an additional 15% to 20% increase in cross-sectional area over conventional sand casting calculations to accommodate the reduced flowability characteristic of lost foam castings.
The pouring temperature was established at 1480°C to ensure adequate fluidity and complete foam decomposition, accounting for the temperature drop of molten metal as it progresses through the mold and degrades the foam pattern. The vacuum pressure was maintained at 0.05 MPa during filling to stabilize the dry sand mold and facilitate removal of foam degradation products. After filling, the vacuum was held for 10 minutes with the pressure reduced to 0.025–0.04 MPa to maintain mold rigidity during solidification, which is particularly beneficial for promoting self-feeding through graphite expansion in ductile iron lost foam castings. The inoculation process involved ladle treatment with a treatment time of 4 minutes, resulting in a maximum Mg treatment time of approximately 12 minutes until pouring completes.
The first trial production run employed a bottom gating system. The gating arrangement was straightforward, consisting of a hollow sprue with a diameter of 45 mm positioned 250 mm above the casting to provide sufficient metallostatic pressure, and a solid foam runner bar with dimensions of 110 mm in length, 50 mm in width, transitioning in height from 40 mm to 10 mm at the ingate. The bottom gating system utilized a self-feeding approach without dedicated feeding risers, incorporating only a small 50 mm × 50 mm × 70 mm skim riser at the top of the upper arm. This design was intended to provide smooth, progressive filling of the cavity from bottom to top, minimizing turbulence and oxidation. However, trial production revealed that although the bottom gating scheme achieved acceptable filling with good surface quality and no wrinkling, sand, or gas defects, severe internal shrinkage defects occurred on the inner wall of the cylindrical section adjacent to the lower arm wall recess. The upper arm region showed acceptable quality, whereas the lower region exhibited major concentrated shrinkage cavities surrounded by extensive shrinkage porosity.
The failure of the first attempt indicated that the self-feeding capacity of the alloy was insufficient to compensate for the isolated hot spots. Consequently, a second trial scheme was developed using a top gating system combined with square blind risers. This design, illustrated in Figure 3, positioned a large central riser above the ingate to delay solidification of the feeding channel, with additional risers placed near major hot spots: at the upper side of the cylinder, the lower side of the cylinder, and the front end of the upper arm. The ingate cross-section was 40 mm × 40 mm, the central top riser was 80 mm × 80 mm × 170 mm, the sprue remained at 45 mm diameter and 250 mm height, and the side risers were 60 mm × 60 mm × 90 mm (upper) and 60 mm × 60 mm × 110 mm (lower), with the front riser at 50 mm × 50 mm × 70 mm. All small holes on the arms were not cast and were subsequently created by machining.
The top gating scheme offered faster filling, a favorable temperature gradient from top to bottom, and improved feeding through the direct connection of risers to the ingate. However, the second trial production results showed that, while the lower part of the casting was free from severe shrinkage defects, serious shrinkage porosity appeared on the inner cylinder wall near the upper arm recess. This pattern suggests that although the upper region received extensive feeding, the heat concentration in the upper hot spot, combined with the cooling of the upper blind riser, led to premature solidification and shrinkage formation in that area.
Numerical Simulation Implementation and Heat Transfer Model Development
The implementation of ProCAST simulation for the lost foam castings involved several stages: geometry creation, mesh generation, material property definition, boundary or initial condition assignment, and solver execution. For geometry creation, I used UG software to create three-dimensional models of the casting, gating system, risers, and sand mold. The components were assembled according to the pouring scheme and exported in Parasolid format for mesh generation. The hollow sprue and pouring cup were modeled as a single empty region, while the casting, risers, and solid foam runner were modeled as a single solid foam region. The sand mold surrounded the gating system and casting.
For mesh generation, I imported the geometry into Visual-Mesh, ProCAST’s integrated mesh generator. The mesh size was controlled with variable density: a smaller element length of 4 mm was used for the casting to accurately capture its complex geometry and thin sections; a larger element length of 15 mm was used for the gating system and risers due to their more regular geometry; and elements of 100 mm were chosen for the sand mold to reduce computational cost. The mesh generation process included geometry cleaning, repair, and automatic tetrahedral element generation. A fine mesh is critical in the ingate region and cast area to ensure accurate flow field resolution, with at least three layers of elements in the ingate cross-section.
For material properties, I initially employed ProCAST’s built-in thermodynamic database to calculate properties for QT600 ductile iron. The calculated liquidus temperature matched the expected value; however, the solidus-to-liquidus interval was unrealistically small at only 6°C. Based on practical knowledge, I adjusted the solidus temperature to 1090°C while retaining the calculated liquidus temperature. The dry sand mold and foam materials were characterized using ProCAST database values. The initial temperature of the sand and foam was set to room temperature (25°C), while the molten metal was prescribed an initial temperature of 1480°C, corresponding to the pouring temperature used in production.
For boundary conditions, heat transfer was modeled using air convection conditions on external surfaces, and a symmetry plane condition was applied to exploit the symmetry of the layout and reduce computational time. For pressure boundary conditions, a vacuum pressure of 0.05 MPa was applied: the pouring cup surface was assigned 1.05 MPa and the external surface of the sand mold was assigned 1.0 MPa. The interface heat transfer coefficient between the casting and the sand mold was initially set to h=150 W/m²·K, which is lower than typical sand casting values to reflect the insulating effect of the refractory coating and dry sand in lost foam castings.
In the thermal-physical model, I defined the interface between the hollow sprue and foam pattern as an “EQUIV” interface to allow continuous heat and mass transfer. Other interfaces were defined as “CONIC” contacts. The effect of the 1.5-mm-thick refractory coating on gas permeability was incorporated by adding corresponding parameters to the data file, as shown in Table 3.
| Parameter | Value | Description |
|---|---|---|
| Coating thickness | 1.5 mm | Refractory coating applied to foam pattern |
| Coating permeability | 1.0 × 10⁻⁹ cm² | Permeability of coating layer |
| Sprue diameter | 45 mm | Hollow foam sprue diameter |
| Pouring temperature | 1480°C | Initial molten metal temperature |
| Vacuum pressure | 0.05 MPa | Negative pressure during filling |
The simulation strategy for lost foam castings required a sequential approach. Because the microstructural calculation requires that no solid fraction exists in the initial state of the cavity, and lost foam castings simulation requires the foam material to be defined as a solid (Empty=No), it is not possible to directly couple thermal, flow, and microstructure models in a single computation as with sand casting. Therefore, I employed a two-step approach: first, the thermal and flow modules were coupled to simulate the filling process, obtaining the temperature distribution at the end of filling; second, this final temperature field was extracted as the initial condition for the subsequent solidification calculation, which coupled the thermal module with the microstructure module. For this specific component with a filling time of approximately 25 seconds per pouring, the temperature loss during filling was considered minimal, allowing the casting temperature at the end of filling to be used directly as the solidification starting temperature.
Simulation Results for the Trial Schemes and Calibration
I performed the filling and solidification simulations for both trial schemes using the established thermal-physical model. The bottom gating scheme simulation results revealed filling characteristics typical of lost foam castings. The metal filled the hollow sprue rapidly in just 0.17 seconds, then proceeded through the foam runner and ingate, with the filling time significantly extending to 1.84 seconds. The presence of the foam pattern markedly reduced the metal flowability. The subsequent filling stages from 20% to 80% cavity fill followed a pattern of approximately 3.5, 2.7, 3.3, 4.8, and 8.8 seconds, with a total filling time of 24.08 seconds, which closely matched the actual measured filling time of approximately 25 seconds. The slight discrepancy is attributable to additional flow resistance from the adhesive used to bond the foam patterns, which was not explicitly modeled in the simulation.
The temperature distribution predicted during filling for the bottom gating scheme showed a relatively stable temperature front, with the last region to fill being located at the furthest distance from the ingate, corresponding to the skim riser position. The final filling location experiences the greatest temperature drop, which consequently diminishes the foam degradation capability, potentially leading to incomplete foam vaporization and entrapment of foam residue. The void fraction analysis indicated that the foam pattern material in the thick-walled area below the cylinder was surrounded by molten metal, leaving the foam in a floating state during degradation. This entrapped foam could potentially form slag or gas defects in the castings if not properly vented. However, the skim riser at the top of the upper arm effectively collected the slag, and the actual castings exhibited good quality in that area.
In the solidification phase, the simulation showed that the skim riser, sprue, and gate solidified early. The regions with the longest solidification time were located at the junctions between the two arm ends and the cylinder, corresponding to the massive thick-walled sections and intersections of wall transitions. These areas developed significant isolated liquid regions or hot spots that could not be fed by the solidified gating system. In the bottom gating scheme, the sprue and ingate solidified prematurely, thereby cutting off the external feeding path. The only remaining feeding mechanism was the self-feeding from graphite expansion. As a result, the simulation predicted the formation of shrinking cavities and porosity concentrated at the thick-walled junctions between the arms and the cylinder, especially in the lower rear region of the cylinder. The predicted cavity shape, size, and position in this area accurately matched the actual trial results. However, the simulation also predicted shrinkage porosity in the upper arm and other hot spots, possibly due to an underestimation of the graphite expansion effect, which overestimates shrinkage.
The top gating scheme simulation showed slightly faster filling at 23.94 s compared to the bottom gating scheme, due to the shorter sprue which reduced initial heat loss. The filling progressed smoothly from top to bottom, with the last region to fill being the lower arm front end. The filling pattern of the top gating scheme demonstrated that the direction of the advancing metal (top-down) opposes the natural upward direction of slag and gas venting, potentially impairing the casting’s self-cleaning ability. Examination of the solidification sequence revealed that the thin cylinder wall solidified relatively early, isolating the upper and lower thick-walled regions. The side risers solidified before the casting hot spots solidified, thus failing to provide adequate feeding. The central large riser delayed solidification of the ingate, offering an extended feeding path to the upper portion of the casting.
The simulation for the top gating scheme using the baseline heat transfer model predicted significant porosity in the lower part of the casting, which contradicted the actual trial results showing the lower part to be sound. This discrepancy highlighted the inadequacy of using a uniform heat transfer setting throughout the sand mold. The internal cavity of the cylinder, being semi-enclosed, experienced a unique thermal environment: the sand inside the cylinder became progressively hotter as filling proceeded, and since it was effectively trapped, the interface between the sand and the inner cylinder wall offered poor heat dissipation, providing insulation rather than cooling. Similarly, the regions around the riser necks experienced concentrated heat accumulation. In the top gating scheme, the lower thick-walled region benefits from multiple feeding paths, including metal from the upper section, feeding from the lower side riser, and self-feeding. The combination of these paths prevents porosity in that area. However, the upper hot spot, despite receiving prolonged feeding from the ingate, is subject to a greater feeding demand because it must also supply metal to the lower sections, thus promoting shrinkage porosity.
To accurately reproduce the complex thermal conditions in these lost foam castings, I modified the heat transfer model by dividing the sand mold into three distinct horizontal zones (upper, middle, and lower) and adjusted the heat transfer coefficients at the casting-mold interfaces accordingly. This approach allowed simulation of the semi-insulating effect of the trapped sand inside the cylinder cavity and the concentrated heat accumulation along the feeder necks. Using this revised heat transfer model, the simulated shrinkage porosity for the top gating scheme closely matched the actual results both in location and size. The validation of the revised heat transfer model through comparison with actual trial results established a reliable thermophysical model for subsequent process improvement studies.
Improved Step Gating System Design and Simulation
Based on the comparative analysis of the two trial schemes and the understanding gained from the simulations, I formulated an improved casting scheme that synthesizes the advantages of both bottom and top gating systems. The design strategy employed a double-layered stepped gating approach where molten metal is introduced simultaneously at both the upper and lower thick-walled sections of the cylinder. This configuration is intended to achieve a more favorable temperature field that promotes directional solidification and ensures good connectivity between the casting and gating system throughout the solidification process. The stepped gating design positions the ingates at the upper and lower hot spots, creating a more uniform temperature distribution and reducing the thermal imbalances that plagued the single-gate designs.
The improved scheme, shown in Figure 5, features a hollow sprue of 45 mm diameter and 250 mm height, connected to a vertical solid foam runner of 60 mm × 60 mm × 240 mm. This runner is strategically connected to the two ingates, serving the dual function of distributing metal and acting as a feeder, with the intention of delaying solidification of the ingates to extend the feeding duration. The ingates measure 60 mm × 40 mm × 40 mm and adopt a variable cross-section at the ingate gates with dimensions of 40 mm × 30 mm to prevent the onset of reverse shrinkage. Two small skim risers, measuring 30 mm × 50 mm × 20 mm each, are placed at the upper side of the cylinder and at the front of the upper arm. These risers are sized to provide slag and gas collection during filling but are designed to solidify within 10 seconds after filling to avoid drawing metal from the casting in the later stages of solidification.
Using the calibrated thermophysical model, I simulated the improved step gating scheme. The simulation predicted a maximum filling time of 16.64 seconds, representing a substantial reduction compared to the 24.08 and 23.94 seconds of the bottom and top gating trial schemes, respectively, due to the dual ingate entry points. The filling process was stable and continuous, with all regions of the pattern completely consumed by the metal, and no significant flow-related defects were predicted. The filling sequence showed that the upper ingate initiated filling before the lower ingate, with a time difference of approximately 2 seconds between the entry of metal from the two layers.
Most importantly, the solidification simulation of the improved scheme showed a significant reduction in the size of the isolated liquid regions at both the upper and lower thick-walled sections. This improvement was attributed to the maintenance of feeder connectivity: the riser and sprue system remained liquid throughout most of the solidification process, enabling continuous external feeding to the hot spots. The simulation predicted the presence of only a small isolated liquid region in the lower part of the casting, which could potentially manifest as minor shrinkage porosity. The porosity model confirmed that the improved scheme eliminated the significant shrinkage cavities predicted in both trial schemes, with significant porosity confined largely to the region of the vertical runner and ingate junction. Based on these encouraging simulation results, I proceeded to validate the improved scheme in actual production.
Production Validation and Further Optimization
The improved step gating scheme was validated in foundry production, and the results of this validation are presented in Figure 6, which displays the foam pattern cluster with the additional strengthening bars added to reduce deformation during handling and sand compaction. The actual production outcomes confirmed the simulation predictions. Sectioning of the castings revealed the absence of significant shrinkage cavities, with only a small amount of concentrated shrinkage porosity present in the lower hot spot region. This result confirmed that the improved scheme effectively resolved the major shrinkage cavity defects that had disabled the trial schemes, while also successfully reducing the severity of shrinkage porosity.
The presence of localized porosity in the lower hot spot drove the subsequent process optimization phase. Based on the earlier investigation into the impact of sand type on shrinkage porosity in ductile iron lost foam castings, I proposed a supplementary measure involving the pre-embedding of chromite sand cores in the wall recess of the lower arm and within the stepped hole of the cylinder. Chromite sand, due to its inherently higher thermal conductivity, exhibits a chilling effect on the adjacent metal, thereby promoting a faster cooling rate and refining the grain structure. This thermal effect creates a steeper thermal gradient that facilitates the feeding of the isolated hot spot by ensuring that the feed path remains open for an extended period. Furthermore, chromite sand possesses magnetic properties, which permit effective separation and reuse through magnetic separation techniques, rendering this approach economically viable. In addition to its thermal benefits, the pre-embedding of chromite sand cores eliminated the risk of loose sand packing in the inconvenient geometric features such as wall recesses and stepped holes, a factor that could introduce variability and defects in castings.
The final production validation, illustrated in Figure 7, confirmed the efficacy of the combined step gating and chromite sand chilling optimization. The resultant castings exhibited a significantly improved internal structure, completely free from macroscopic shrinkage cavities and with shrinkage porosity minimized to an acceptable level. The castings satisfied the stringent quality requirements for strength and internal integrity, thereby achieving the objective of producing defect-minimized lost foam castings in a cost-effective and reproducible manner.
I systematically documented the connection between the evolution of the gating scheme and the corresponding simulation predictions throughout this investigation, and the comparisons are summarized in the following table, which compares the filling and defect characteristics of the different schemes:
| Scheme | Filling Time (s) | Filling/Mold Stability | Slag/Gas Removal | Feeding Mechanism | Shrinkage Outcome |
|---|---|---|---|---|---|
| Bottom gating (trial 1) | 24.1 | Smooth, stable | Skim riser only, adequate | Self-feeding only | Severe cavities in lower inner wall |
| Top gating (trial 2) | 23.9 | Smooth, stable | Opposed, poor | Central riser + side risers | Severe porosity in upper inner wall |
| Step gating (improved) | 16.6 | Smooth, stable | Small skim risers, adequate | Runner-feeder with delayed solidification | No major cavities; minor lower porosity |
| Step gating + chromite sand | — | Smooth, stable | Small skim risers, adequate | Runner-feeder + chilling | Defect-minimized casting |
The systematic application of the calibrated thermal-physical model allowed me to achieve a high degree of predictive accuracy, thereby enabling the transition from empirical “trial-and-error” methods to simulation-driven process design. The research conclusively demonstrates that when an accurate thermophysical model is established through careful parameter calibration and validation, numerical simulation can serve as a powerful tool for optimizing the lost foam castings process, identifying potential defect locations, and guiding the design and refinement of pouring systems. This approach dramatically reduces the number of physical trials required, conserves materials and labor, and minimizes the substantial costs associated with traditional development methodologies.
Conclusions and Future Perspectives
This dissertation has systematically examined the lost foam casting process for the production of ductile iron lift arms, focusing on the development of a robust simulation methodology and the application of that methodology to optimize the casting process design. The primary conclusions arising from this work are summarized as follows:
First, the investigation established the special simulation parameter settings required for accurately modeling the lost foam casting process and the shrinkage porosity formation in ductile iron castings. Key parameters such as the foam heat transfer coefficients, gas models, coating permeability, and the parameters controlling graphite expansion in the microstructure model were evaluated through comparative simulations with actual casting trials. It was found that the proper representation of the thermal properties of the sand mold, particularly the semi-insulating effect of entrapped sand in internal cavities, is crucial for accurate shrinkage porosity predictions in complex geometries. Moreover, the graphite expansion parameters, including GRAPHITE, FADING, MGTREAT, and MOLDRIG, have a significant influence on the simulated shrinkage trend, emphasizing the need for careful calibration based on actual production conditions.
Second, the research compared the impact of different sand materials on the solidification behavior of ductile iron castings. The simulation results demonstrated that using chromite sand with its elevated thermal conductivity induces a chilling effect, accelerating the cooling rate sufficiently to reduce shrinkage porosity and promote a more dispersed distribution of residual porosity, a finding that directly supported the final optimization strategy.
Third, the successful application of ProCAST simulation to the lost foam castings process for complex three-dimensional components was demonstrated. By employing a two-step approach of coupled thermal-flow simulation for filling followed by a coupled thermal-microstructure simulation for feeding and solidification, I was able to model the filling and solidification sequence of the lift arms accurately. This methodology serves as a general guideline for applying ProCAST to simulate lost foam castings manufacturing processes.
Fourth, the research formulated a rational gating system design, evolving from a simple bottom gating system to a top gating system and ultimately to a double-layered stepped gating system. The simulation played an essential role in developing the improved system designs by predicting, refining, and validating the behavior of each scheme. The combination of the stepped gating system with pre-embedded chromite sand chilling cores constituted a final process that effectively minimized internal shrinkage defects in the ductile iron lift arms.
Looking toward the future, this study underscores the significant potential of computer simulation as a foundational tool for the advancement of lost foam casting technology. Future work should focus on the development of a comprehensive property database for foam, coatings, and sands specifically tailored to the lost foam process. Further studies should expand the scope of the simulation to include stress and deformation analyses for predicting distortion in castings and to incorporate microstructural evolution models for linking process parameters to final mechanical properties. Additionally, the broader adoption of such simulation tools in small and medium-sized foundries requires the creation of more user-friendly interfaces and simplified calibration procedures that lower the barrier to entry for the industrial workforce. This research contributes to the growing body of knowledge supporting the deployment of computer simulation in process design for lost foam castings, an indispensable step toward achieving high-quality, cost-effective, and sustainable production.