In our foundry operations, we frequently encountered premature failure of welded steel water-ring vacuum pump housings due to erosion from water and sand particles. Leakage led to reduced system pressure, impacting overall casting production efficiency and quality. To resolve this, we initiated a project to produce these housings using ductile iron castings, leveraging their superior wear resistance, pressure tightness, and durability. The target component was a large, thick-section housing with demanding technical requirements.
The vacuum pump housing is a significant casting. Its key dimensions are an outer diameter of 864 mm, a height of 637 mm, and a nominal wall thickness of 35 mm in the cylindrical section. The as-cast weight is approximately 425 kg. The specified material is QT600-3 (equivalent to ASTM A536 60-40-18 or EN-GJS-600-3), with the paramount technical requirement being pressure tightness—the casting must be free from leaks. The production of such thick-section ductile iron castings requires meticulous control over the entire process to prevent defects like shrinkage porosity, slag inclusions, and dimensional inaccuracies.
The success of high-integrity ductile iron castings hinges on a holistic approach encompassing pattern making, molding, melting, and pouring. For this project, we selected the Lost Foam Casting (LFC) process due to its advantages in producing complex geometries with high dimensional accuracy and excellent surface finish, which is ideal for near-net-shape components like pump housings.
1. Foundry Method Engineering and Simulation
The initial and most critical phase was designing a robust casting method. We employed a three-dimensional CAD model to design the gating and feeding system. Given the component’s size and weight, we opted for a one-casting-per-mold configuration using a flask measuring 1200 mm × 1000 mm × 1300 mm. The total poured weight, including the gating system and feeders, was calculated to be 560 kg, requiring a 700 kg capacity treatment ladle.
To ensure complete filling and soundness in these ductile iron castings, we designed a top-gating system with a semi-open choke ratio. The principle is to establish a controlled, progressive fill from the bottom upwards and ensure adequate feeding during solidification. The designed gating ratio was:
$$ \Sigma F_{\text{sprue}} : \Sigma F_{\text{runner}} : \Sigma F_{\text{ingate}} = 1 : 1.4 : 1.2 $$
The cross-sectional areas were:
| Element | Description | Dimensions | Cross-Sectional Area (mm²) |
|---|---|---|---|
| Sprue | Vertical channel | Ø50 mm | ≈1963 |
| Runner | Horizontal distribution | 75 mm × 75 mm (square) | 5625 |
| Ingate | Entry to cavity | 40 mm × 140 mm (rectangular) | 5600 |
Feeding is paramount for thick-section ductile iron castings to counteract the shrinkage associated with the austenitic solidification. Instead of external chills (difficult to place in LFC) or internal chills (risk of poor fusion), we employed six feeding risers positioned on the top flange of the housing. Two of these risers also served as hot in-gates. The concept involves molten metal entering through these hot risers, dropping to the bottom of the cavity, and then filling upward in a calm, controlled manner. Once the metal reaches the top, it fills the remaining four risers. All six risers (120 mm × 120 mm × 140 mm) then act as liquid reservoirs to feed the solidifying casting, promoting directional solidification from the casting extremities toward the risers.
To validate this design, we performed numerical simulation using MAGMAsoft. The filling and solidification sequences were analyzed to check for turbulence, cold shuts, and potential shrinkage zones.
Filling Simulation Results: The simulation confirmed a smooth, bottom-up filling sequence. Metal entering from the top gates fell directly to the bottom of the cavity without excessive splashing or surface turbulence. The metal front then rose steadily, with the cooler metal at the bottom being reheated and fed by the hotter metal above it during the fill. This minimizes premature solidification at the base.
Solidification Simulation Results: The solidification pattern showed a clear progression from the thin outer walls and bottom towards the thicker top sections and finally the risers. The risers remained liquid longest, fulfilling their intended function as feeders. The simulation predicted no major isolated hot spots or macro-shrinkage cavities. Any micro-shrinkage potential in isolated liquid pockets was deemed acceptable, as the graphite expansion characteristic of ductile iron castings can often compensate for this.
2. Lost Foam Pattern Production and Assembly
The size of the housing made it impossible to machine the expendable pattern (white pattern) from a single block of expanded polystyrene (EPS). We, therefore, adopted a segment-and-assemble strategy.
First, the 3D CAD model of the housing was decomposed into manageable segments. The cylindrical body was split into 12 curved sections, and each end flange was divided into 8 segments. Internal radii (R10 mm) were created using thin, pre-curved foam strips. This decomposition was programmed for a CNC hot-wire cutting machine to ensure dimensional accuracy and repeatability for each segment.
The assembly was a meticulous manual process. We began by establishing a centerline on a flat glass table. Foam segments were then carefully aligned and temporarily fixed using fiberglass rods and a low-residue adhesive. All seams between segments were filled with specialized gap-filling paste and sealed with paper tape to prevent coating penetration during the subsequent dipping process. To prevent distortion of the large, thin-walled pattern during coating and sand filling, we reinforced the structure with four wooden braces (15 mm × 15 mm) attached to the top and bottom flanges. Finally, the gating system (sprue, runner, ingates) and the six risers were assembled and attached to the main pattern using fiberglass rods for added strength. The completed cluster was robust enough to withstand the rigors of the coating and molding processes.
3. Coating, Molding, and Pouring Parameters
The ceramic coating applied to the foam pattern is critical in LFC. It creates a barrier between the degrading foam and the sand, controls heat transfer, and ensures surface finish. For these thick-section ductile iron castings, we required a coating with high refractoriness and permeability.
The assembled pattern cluster was dipped four times in a zirconia-based refractory slurry to build a sufficient coating thickness. The slurry density was strictly controlled between 1.53 to 1.55 g/cm³ (approximately 69–71 °Bé). After each dip, the pattern was drained and dried in a temperature- and humidity-controlled oven to remove moisture completely. The final “yellow pattern” was fully sealed and rigid.
For molding, the yellow pattern was placed in the flask on a leveled bed of dry, unbonded silica sand. The flask was then filled gently with more sand while being subjected to three-dimensional vibration for 90 seconds. This ensures the sand flows into all cavities and packs uniformly around the pattern, providing excellent support.
The pouring parameters were set based on simulation and experience with similar ductile iron castings:
| Parameter | Value | Purpose |
|---|---|---|
| Pouring Temperature | 1430 – 1450 °C | Ensures fluidity, complete foam degradation, and adequate feeding. |
| Mold Vacuum | -0.05 to -0.06 MPa | Removes foam pyrolysis gases, stabilizes the mold, and prevents sand collapse. |
| Pressure Holding Time | 60 minutes after pour | Maintains mold integrity until the casting has solidified sufficiently. |
To guard against accidental superheating, clean steel chill blocks were kept on hand to cool the metal in the ladle if the temperature exceeded 1450°C. The sprue was topped up during pouring to maintain a metallostatic head for feeding.
4. Melt Processing and Cored Wire Inoculation
The chemical composition is fundamental to achieving the required microstructure and mechanical properties in ductile iron castings. Our target composition for QT600-3 is shown below:
| Element | Target Range (wt.%) | Function/Rationale |
|---|---|---|
| Carbon (C) | 3.6 – 3.8 | Promotes graphitization, improves fluidity and castability. |
| Silicon (Si) – Final | 2.3 – 2.5 | Strong graphitiser; controls matrix ferrite/pearlite ratio. |
| Manganese (Mn) | 0.4 – 0.6 | Strengthens pearlite; kept low to minimize segregation. |
| Sulfur (S) | < 0.015 | Must be very low for successful nodularization. |
| Phosphorus (P) | < 0.05 | Kept low to prevent phosphide eutectic, which embrittles. |
| Magnesium (Mg) – Residual | 0.04 – 0.06 | Essential for spheroidal graphite formation. |
| Tin (Sn) | ~0.06 | Pearlite stabilizer to ensure strength in thick sections. |
We employed a highly efficient and reproducible treatment process: the Tundish Cover Ladle method with dual-wire inoculation. The process sequence is as follows:
1. Base Iron: Iron is melted in a coreless induction furnace to a temperature above 1500°C with low sulfur content.
2. Ladle Pre-treatment: 0.3% of a FeSi-based pre-inoculant is added to the empty treatment ladle to prepare the iron for nodularization.
3. Nodularization: The ladle is filled with base iron. A twin-wire feeder injects cored wire containing FeSiMg25RE3 alloy directly into the bottom of the molten metal stream beneath a tundish cover. The addition rate of nodularizer is precisely 0.7% of the iron weight. The cover and bottom injection minimize Mg oxidation and fuming, leading to high and consistent Mg recovery (typically >70%). The reaction is calm and controllable.
4. Post-Inoculation: Immediately after wire feeding, 0.2% of a foundry-grade 75% FeSi alloy is added as a post-inoculant to enhance graphite nucleation and prevent chill.
5. Slag Removal & Pouring: Slag is thoroughly skimmed, and the treated iron is transferred to the pouring station.
The precise wire feeding is governed by a PLC, controlling speed and length according to the equation:
$$ L_w = \frac{{m_{iron} \times \%_{MgAdd}}}{{LC_w \times \rho_w}} $$
Where \( L_w \) is the wire length to feed (m), \( m_{iron} \) is the mass of iron (kg), \( \%_{MgAdd} \) is the target magnesium addition percentage, \( LC_w \) is the linear content of Mg in the wire (kg/m), and \( \rho_w \) is a parameter accounting for recovery. This ensures repeatable treatment for every batch of ductile iron castings.
5. Results, Inspection, and Conclusion
The castings were shaken out, shot-blasted, and the gating/riser systems were removed. Visual inspection of the riser necks showed sound, dense metal with no evidence of gross shrinkage pipes or cavities. Dimensional checks confirmed the high accuracy afforded by the LFC process, with all critical dimensions within the drawing specifications.
Test coupons attached to the casting were used for mechanical and metallurgical evaluation. The results confirmed that the ductile iron castings met the QT600-3 specification:
| Property | Result | Specification (Typical for QT600-3) |
|---|---|---|
| Tensile Strength (Rm) | 641 MPa | > 600 MPa |
| Yield Strength (Rp0.2) | 410 MPa | > 370 MPa |
| Elongation (A) | 3.5 % | > 3 % |
| Hardness (HBW) | 220-240 | 190-270 |
| Nodularity | > 90% | > 80% |
| Nodule Count | 120-140 nodules/mm² | -> 100 nodules/mm² |
| Matrix Structure | ~65% Pearlite, balance ferrite | Predominantly pearlitic |
Subsequent machining of the pilot castings revealed no subsurface defects such as shrinkage porosity or sand/slag inclusions in the pressure-bearing walls. The final component underwent a hydrostatic pressure test and passed without any leakage, fully satisfying the service requirement.
6. Summary and Key Learnings
This project successfully demonstrated the viability of producing large, thick-section, pressure-tight ductile iron castings using the Lost Foam Casting process. The integration of modern design, simulation, and processing tools was crucial.
1. Process Feasibility: The Lost Foam process is exceptionally well-suited for complex, near-net-shape components like pump housings. The segmentation of the white pattern via CNC cutting and its manual assembly allowed for the production of a large pattern with high dimensional fidelity, making it ideal for low-to-medium volume production runs of ductile iron castings.
2. Melt Treatment Control: The adoption of the tundish cover ladle with dual-wire inoculation provided a clean, efficient, and highly controllable method for nodularizing and inoculating the iron. The PLC-controlled injection guaranteed consistent treatment quality, which is the foundation for achieving the required microstructure and mechanical properties in these ductile iron castings. The process minimized variability and environmental emissions compared to traditional sandwich or converter ladle treatments.
3. Method Design Efficacy: The casting method, featuring a top-gating system combined with strategically placed risers, was validated by simulation and practical results. It facilitated a tranquil fill pattern and promoted directional solidification. The final castings were dense, free from macro-shrinkage, and met the stringent pressure-tightness criterion. The design effectively harnessed both the liquid feeding from the risers and the graphitic expansion inherent to ductile iron castings to achieve soundness.
In conclusion, the successful production of the vacuum pump housing underscores that a systematic approach—combining advanced pattern engineering, rigorous process simulation, controlled melt treatment, and precise pouring practices—is essential for manufacturing high-integrity, thick-section ductile iron castings for demanding applications.