In the production of complex structural components such as cast-steel joints for steel frame buildings, the conventional sand casting foundry approach has long faced challenges: high pattern costs, difficulty in mold parting, and significant waste of wood for single-piece or small-batch production. Moreover, the stringent requirements for weldability, toughness, and dimensional accuracy make the cast-steel joint a demanding product. To address these issues, our team at the Materials Science and Engineering School developed a composite process that combines the design freedom of lost foam with the reliability of traditional sand casting foundry techniques. This article presents the methodology, process optimization, and numerical simulation results that have been successfully implemented in actual production.
The principle of the composite lost foam–sand casting foundry method is straightforward. Instead of using an entire wood pattern, we replace the external contour of the casting with a foam pattern. After the mold is made with water glass sand, the foam is destructively removed, leaving a hollow cavity. Then, pre‑made sand cores (produced from wood patterns) are placed inside, and the mold is closed for pouring. This approach eliminates the need for complex parting lines and undercut handling typical in conventional sand casting foundry operations. It also avoids the carbon contamination problem inherent in full lost foam casting of steel, because the foam is removed before metal enters the cavity.
1. Process Overview and Advantages
The composite process can be broken down into the following steps:
- 3D Modeling and Foam Pattern Fabrication: The joint geometry is modeled in CAD software. Cavities (holes or hollow sections) are identified and will be formed by subsequent sand cores. The solid model (with cavities filled) is sliced into manageable segments. Each segment is cut from low‑density polystyrene (EPS) foam using CNC or manual cutting, then glued together to form the full external pattern.
- Sand Core Preparation: For each internal cavity, a wooden pattern is used to mold a water‑glass sand core. The core is cured with organic ester hardener, providing sufficient strength and collapsibility.
- Molding with Water‑Glass Sand: The foam pattern is coated with an alcohol‑based refractory paint (minimum thickness 2 mm) and placed in a flask. Water‑glass sand (sodium silicate bonded) with organic ester hardener is filled and compacted. The sand casting foundry typically uses vibration to ensure density.
- Foam Removal and Core Setting: After the sand mold has fully hardened, the foam pattern is mechanically broken and removed from the cavity. Then the pre‑made sand cores are inserted into the hollow spaces. The cores are accurately located by core prints that extend beyond the casting surface.
- Pouring and Solidification: The mold is closed and poured with liquid steel (e.g., GS‑20Mn5, a weldable structural steel grade). Because the cavity is empty, no foam decomposition gases are generated, eliminating carbon pickup and gas defects.
- Post‑Processing: After cooling, the casting is shake‑out, cores are removed, and gates/risers are cut. Heat treatment (normalizing + tempering) relieves residual stresses. Finally, shot blasting, grinding, and anti‑corrosion coating are applied.
The composite method retains the advantages of the sand casting foundry: reliable mechanical properties, flexible alloy selection, and lower capital investment compared to full lost foam. At the same time, it gains the geometric freedom of foam patterns, reducing pattern cost by 30%–50% for complex joints and shortening lead time by about 40%.
2. Key Equations for Solidification Simulation
To ensure sound castings without shrinkage porosity or hot spots, we employed the Huazhu CAE system (InteCAST) for numerical simulation. The governing equation for heat transfer during solidification is the Fourier‑Kirchhoff equation with a latent heat source term:
$$ \rho c \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \rho L \frac{\partial f_s}{\partial t} $$
where:
- \( \rho \) = density of the steel (kg/m³)
- \( c \) = specific heat capacity (J/(kg·K))
- \( k \) = thermal conductivity (W/(m·K))
- \( T \) = temperature (K)
- \( t \) = time (s)
- \( L \) = latent heat of fusion (J/kg)
- \( f_s \) = solid fraction (0 ≤ f_s ≤ 1)
The solid fraction evolution is approximated by the Scheil equation for a binary alloy:
$$ f_s = 1 – \left( \frac{T_m – T}{T_m – T_l} \right)^{\frac{1}{k_0 – 1}} $$
where \( T_m \) is the melting point of pure iron, \( T_l \) is the liquidus temperature, and \( k_0 \) is the equilibrium partition coefficient. For the GS‑20Mn5 steel used (0.20% C, 1.5% Mn), typical values are:
| Property | Value | Unit |
|---|---|---|
| Liquidus temperature | 1515 | °C |
| Solidus temperature | 1440 | °C |
| Latent heat | 270 | kJ/kg |
| Thermal conductivity (solid) | 30 | W/(m·K) |
| Thermal conductivity (liquid) | 25 | W/(m·K) |
| Specific heat (solid) | 700 | J/(kg·K) |
| Specific heat (liquid) | 800 | J/(kg·K) |
| Density | 7800 | kg/m³ |
The simulation was run using the finite‑difference method on a uniform grid (mesh size 5 mm). The total number of elements for the largest joint (mass 22 140 kg, maximum tube diameter 1080 mm) was approximately 1.2 million.
3. Case Study: A Multi‑Tube Crown‑Valley Joint
We applied the composite process to a typical complex joint used in the main stadium of a large sports event. The joint has six intersecting tubes with diameters ranging from 600 mm to 1080 mm. Wall thickness at the intersection is 60 mm. The total mass is 22 140 kg.
3.1 Foam Pattern Design
From the 3D model, we first extracted the intersection curves of the six tubes. The solid model (with all cavities filled) was divided into 12 foam blocks along the parting lines. Each block was cut from cylindrical EPS foam (density 20 kg/m³) using a hot‑wire cutter guided by coordinates measured from the CAD model. The blocks were glued together with a solvent‑based adhesive. The final foam pattern weighed approximately 45 kg (about 1/500 of the casting weight).
3.2 Gating and Risering System
The gating system was designed to ensure directional solidification and smooth filling. The layout consisted of:
- One main sprue (diameter 120 mm) placed at the highest point of the casting.
- A bottom runner bar connecting all six tube ends.
- Six ingates (one per tube) located at the lowest point of each tube to promote bottom filling and avoid turbulence.
- One main riser (diameter 350 mm, height 500 mm) placed on the largest thermal center. Additionally, each tube had a small side riser (diameter 150 mm) to assist feeding.
The pouring temperature was 1550 °C, and the pouring time was calculated from the critical velocity to avoid sand erosion:
$$ t_{pour} = \frac{m}{\rho \cdot A_{choke} \cdot v} $$
where \( m \) is the total mass of steel (including risers, about 28 t), \( A_{choke} \) is the choke area (total ingate area of 12 000 mm²), and \( v \) is the average flow velocity (≈ 0.5 m/s). This gave a pouring time of about 90 seconds.
| Parameter | Value |
|---|---|
| Number of ingates | 6 |
| Ingate cross‑section (each) | 2000 mm² |
| Total ingate area | 12 000 mm² |
| Main riser volume | 0.048 m³ |
| Side riser volume (each) | 0.007 m³ |
| Pouring temperature | 1550 °C |
| Mold temperature (preheat) | 200 °C |
3.3 Numerical Simulation Results
Using Huazhu CAE, we simulated the filling and solidification. The filling simulation showed non‑turbulent flow with no cold shuts. The temperature distribution at the end of filling indicated that the risers remained the hottest zones, ensuring effective feeding.
The solidification progress was monitored by tracking the liquid fraction over time. The entire casting solidified in about 48 663 s (≈ 13.5 h). The riser was the last to solidify, confirming directional solidification from the thinner walls toward the thermal centers. No isolated liquid islands were detected, meaning that shrinkage porosity was avoided. The simulation predicted less than 0.5% porosity by volume, which was within the acceptance criteria for the project.
We also computed the temperature gradient (\( G \)) and solidification rate (\( R \)) at the hot spots. The Niyama criterion (\( G/\sqrt{R} \)) was used to evaluate shrinkage risk:
$$ \text{Niyama} = \frac{G}{\sqrt{R}} $$
where \( G \) is the thermal gradient (K/mm) and \( R \) is the cooling rate (K/s). For steel, a Niyama value above 1 K⁰.⁵·mm⁻¹·s⁻⁰.⁵ generally indicates a sound casting. In our simulation, the minimum Niyama value was 1.8 K⁰.⁵·mm⁻¹·s⁻⁰.⁵, confirming robust feeding.
4. Experimental Validation and Production Data
Three prototype joints were produced using the composite process in our sand casting foundry. Table 3 summarizes the key production indicators compared with conventional wood‑pattern water‑glass sand casting.
| Parameter | Conventional wood pattern | Composite (foam+core) |
|---|---|---|
| Pattern cost (per joint) | ~ $15 000 (wood) | ~ $4 500 (foam + wood cores) |
| Pattern production time | 6 weeks | 2 weeks |
| Mold making time (per casting) | 3 days | 2 days |
| Dimensional accuracy | ±2 mm (critical) | ±1.5 mm (critical) |
| Carbon contamination risk | None | None (foam removed) |
| First‑pass yield (non‑destructive testing) | 75% | 92% |
| Overall production lead time | 12 weeks | 6 weeks |
The composite process reduced pattern cost by 70% and cut total lead time in half. The yield improvement from 75% to 92% was attributed to the more consistent mold cavity (no wood pattern distortion) and the optimized gating system designed via simulation.
Chemical analysis of the final castings met the GS‑20Mn5 specification: C 0.18–0.22%, Si 0.30–0.60%, Mn 1.20–1.60%, P ≤ 0.020%, S ≤ 0.015%. Tensile tests gave yield strength 300 MPa, ultimate tensile strength 500 MPa, elongation 22%, and impact energy (‑20 °C) 40 J, all exceeding requirements. No significant carbon gradient was observed, confirming the absence of carburization from foam residues.
5. Discussion on the Role of the Sand Casting Foundry
The success of this composite method heavily relies on the capabilities of a modern sand casting foundry. Water‑glass sand with organic ester hardener provides excellent collapsibility, which is vital for removing the foam pattern without damaging the mold cavity. The sand casting foundry must also control the sand grain size (AFS 50–60) and the binder content (typically 3–4% sodium silicate and 0.3–0.5% organic ester) to achieve proper strength and permeability. In our trials, the sand mold compressive strength after 24 h was about 1.5 MPa, sufficient to withstand the ferrostatic pressure of 28 tons of liquid steel.
Furthermore, the sand casting foundry equipment—vibration tables, mixing machines, and ladle preheaters—must be adapted for the foam‑removal step. We developed a simple mechanical claw tool to break and extract the foam in less than 30 minutes. The hollow cavity is then inspected by borescope for any residual foam fragments, which are removed by compressed air.
The composite process also reduces the dependency on skilled patternmakers. In a conventional sand casting foundry, making complex wood patterns for joints with multiple intersecting tubes requires exceptional craftsmanship and months of work. With foam, the design is entirely digital, and the cutting can be outsourced or performed by CNC routers. This democratizes the production of complex steel castings, making it accessible to smaller sand casting foundries that may lack experienced wood pattern makers.
6. Thermal Stress and Hot Tearing Considerations
One concern with large steel castings is hot tearing, especially at thick‑thin transitions. We used the simulation to compute the thermal stress field during solidification. The stress‑strain relationship was modeled using an elasto‑viscoplastic constitutive law with temperature‑dependent Young’s modulus and yield strength:
$$ \sigma = E(T) \cdot (\varepsilon – \varepsilon_{th} – \varepsilon_p) $$
where \( \varepsilon_{th} \) is the thermal strain, \( \varepsilon_p \) the plastic strain, and \( E(T) \) the temperature‑dependent modulus. For GS‑20Mn5, at temperatures above 1200 °C, the modulus drops rapidly, and the material behaves almost viscously. To avoid hot tears, we ensured that the critical strain for hot tearing did not exceed 0.5% at any point. The simulation showed maximum strain of 0.35% in the intersection areas, which is acceptable. Post‑casting inspection (dye penetrant and ultrasonic) confirmed zero hot tears in the three prototypes.
7. Economic and Environmental Benefits
Beyond cost and lead time, the composite process offers environmental advantages. Wood patterns, especially for large joints, require large logs that are often discarded after one use. Foam patterns, although made from polystyrene, are much lighter (1/500 of casting weight) and generate less solid waste. Moreover, the foam removal does not involve burning, so no harmful fumes are emitted inside the sand casting foundry. The water‑glass sand can be reclaimed by mechanical or chemical regeneration, further reducing the environmental footprint.
In terms of energy consumption, the elimination of complex wood patterns reduces the energy needed for pattern storage and handling. The overall energy per ton of good casting was estimated to be 8% lower compared to the conventional route, mainly due to a higher yield (less remelting of defective castings).
8. Limitations and Future Work
While the composite method has proven successful for the specific joint geometry described, some limitations exist:
- The foam pattern must be sufficiently rigid to survive the water‑glass sand compaction process. For very thin walls (<20 mm) or large unsupported spans, the foam may deform. In such cases, we reinforce the pattern with internal foam ribs that are later removed.
- The destructive removal of foam creates debris that must be fully evacuated. For cavities with complex internal geometries, blind spots may retain foam particles. We are developing a vacuum‑assisted removal system combined with chemical dissolution (using a bio‑based solvent) for future applications.
- The organic ester‑hardened water‑glass sand has a limited bench life (about 20–30 minutes at 25 °C). The sand casting foundry must carefully schedule the mixing and molding sequence to avoid premature hardening before the foam is fully placed.
We are currently extending the composite approach to other complex steel castings, such as pump impellers, valve bodies, and ship propellers. Preliminary results for a 12‑ton impeller casting show similar improvements in yield and dimensional accuracy.
9. Conclusion
The composite lost foam and sand casting foundry process for cast‑steel joints represents a significant advance over traditional wood‑pattern methods. By combining the geometric flexibility of foam with the reliability of water‑glass sand molding, we achieved a 70% reduction in pattern cost, 50% shorter lead time, and a 17% increase in first‑pass yield. Numerical simulation using Huazhu CAE played a critical role in optimizing the gating and risering system, ensuring directional solidification and defect‑free castings. The process has been successfully implemented in a production sand casting foundry and can be readily adopted for other complex ferrous castings.
In summary, the sand casting foundry that embraces digital foam patterns and selective core integration can compete effectively with other near‑net‑shape processes while retaining the material versatility and low capital cost that make sand casting foundry the backbone of heavy‑section steel production.