The failure of a welded steel water-ring vacuum pump housing in our foundry’s production line presented a significant operational challenge. The continuous erosion from water and sand particles led to leaks, causing a drop in system pressure that directly impacted casting quality and productivity. To resolve this, we initiated a project to redesign and produce the housing using nodular cast iron. The superior wear resistance, pressure tightness, and castability of nodular cast iron made it an ideal candidate for this demanding application. This document details the first-person perspective of the complete process development, from initial design and simulation to successful production, for this complex, thick-section casting.
The target component was a large vacuum pump housing with major dimensions of φ864 mm in diameter and 637 mm in height. The cylindrical body had a nominal wall thickness of 35 mm, and the final casting weight was specified at 425 kg. The material specification was QT600-3 (equivalent to ASTM A536 80-55-06), with the critical technical requirement being pressure tightness—the casting must be free from leaks. This requirement immediately directed our focus towards achieving a sound, shrinkage-free microstructure, which is paramount for nodular cast iron components in pressure service.
Casting Process Design and Rationale
Given the size and complexity of the part, the lost foam process was selected for its ability to produce precise, intricate shapes without the need for cores, minimizing parting lines and potential leak paths. A one-casting-per-flask configuration was planned using a standard flask measuring 1200 mm × 1000 mm × 1300 mm. The total weight of sand and pattern was estimated at 560 kg. To ensure adequate molten metal, a 700 kg treatment ladle was designated.
The core of the casting method involved a top-gating system combined with multiple risers. A semi-open gating system was designed with a choke at the sprue base to control initial turbulence. The designed gating ratio was ΣFsprue : ΣFrunner : ΣFingate = 1 : 1.4 : 1.2. The sprue diameter was set at φ50 mm. The cross-sectional areas were calculated as follows:
$$ A_{sprue} = \pi \times (25)^2 \approx 1963.5 \, mm^2 $$
$$ A_{runner} = 1.4 \times A_{sprue} \approx 2749 \, mm^2 \quad \text{(Designed as 75mm x 75mm = 5625 mm$^2$ total for two runners)} $$
$$ A_{ingate} = 1.2 \times A_{sprue} \approx 2356 \, mm^2 \quad \text{(Designed as 40mm x 140mm = 5600 mm$^2$ total)} $$
Six risers, each 120 mm × 120 mm × 140 mm, were evenly distributed on the top surface of the housing. Two of these risers served as hot risers, where the metal entered the cavity. The logic was that the iron would first drop to the bottom of the cavity, then fill progressively upward. Once the cavity was full, the metal would rise into the remaining four risers. All six risers would then act as feeding sources during solidification, compensating for the volumetric shrinkage of the nodular cast iron. The use of external chills was impractical in the loose sand of the lost foam process, and internal chills were avoided due to risks of poor fusion.
| Parameter | Value/Specification |
|---|---|
| Casting Material | QT600-3 (Nodular Cast Iron) |
| Casting Weight | 425 kg |
| Flask Size | 1200 x 1000 x 1300 mm |
| Pouring Temperature | 1430 – 1450 °C |
| Gating System Type | Top Gating with Riser Feeding |
| Vacuum Level | -0.05 to -0.06 MPa |
| Pattern Hold Time After Pouring | 1 hour |
Numerical Simulation for Process Validation
Before committing to physical production, a MAGMAsoft numerical simulation was conducted to validate the filling and solidification behavior. The primary concerns were to ensure smooth, non-turbulent filling to prevent slag entrainment and to confirm that the risers would effectively feed the entire casting, eliminating macro-shrinkage.
The filling simulation confirmed the design intent. The metal entered through the hot risers, streamed down to the bottom of the cavity with minimal splashing, and then filled the mold in a steady, upward progression. This controlled fill helped to maintain thermal gradients favorable for directional solidification. The simulation snapshots showed no premature freezing of in-gates or undesirable fountain effects.
The solidification simulation was even more critical. The results indicated that the thermal center of the casting, particularly in the thick cylindrical sections, remained liquid longest. The six risers on top solidified last, creating the necessary pressure gradient to draw feed metal from the risers into the casting. The simulation predicted only minor, isolated liquid pools in heavy sections, which are typically counteracted by the graphite expansion phase inherent to nodular cast iron solidification. The absence of major isolated hot spots confirmed the adequacy of the risering scheme. The solidification sequence can be conceptually summarized by Chvorinov’s rule, where the riser solidification time ($t_{riser}$) must be greater than the casting solidification time ($t_{casting}$):
$$ t = B \left( \frac{V}{A} \right)^n $$
where $V$ is volume, $A$ is surface area, $B$ is a mold constant, and $n$ is an exponent (often ~2). Our design ensured $(V/A)_{riser} > (V/A)_{critical-casting-section}$.
Pattern Production and Assembly
Producing the expansive white pattern for the pump housing as a single piece was not feasible. The pattern was therefore strategically decomposed into multiple segments using CAD software. The large cylindrical body was split into 12 slices, while the two end flanges were divided into 8 segments total. This segmentation allowed for cutting from standard foam block sizes and facilitated easier handling and gluing. Sharp corners from cutting were rounded to a radius of R10 mm using specialized foam fillet sticks.
The assembly was a meticulous, manual process. It began on a flat glass plate with a pre-marked centerline to ensure dimensional accuracy. Foam segments were carefully aligned and temporarily fixed using low-fume hot glue and fiberglass reinforcing rods. All seams were then filled with a specialized foam gap-filler adhesive and wrapped with paper tape to create a perfectly seamless surface, preventing coating penetration. To combat potential distortion during the coating process—a common issue with large foam patterns—four reinforcing wooden strips (15 mm x 15 mm) were glued to the top and bottom of the assembly. Finally, the sprue, runners, and risers were attached and reinforced. The completed pattern assembly was robust and dimensionally precise.
Coating, Molding, and Pouring
A refractory coating is essential in lost foam casting to prevent sand erosion and ensure a good surface finish. A zirconia-based coating with a controlled viscosity (69-71 °Bé) was selected. The large pattern required careful dipping and rotating to achieve full, uniform coverage. This coating and drying cycle was repeated four times to build a sufficient barrier layer thickness. The coated “brown pattern” was then thoroughly dried in a controlled oven to remove all moisture, which is crucial to avoid gas defects during pouring.
For molding, the dried pattern was placed in the flask on a leveled sand base. Dry, unbonded silica sand was then poured around it while the flask was subjected to 3D vibration for 90 seconds. This ensures the sand fully and tightly fills all cavities around the pattern without damaging it. The gating system was connected to the pouring cup, and the flask was placed under a vacuum of -0.05 to -0.06 MPa.
Metal Treatment: The Heart of Nodular Cast Iron Production
The metallurgical treatment is the most critical step in producing high-quality nodular cast iron. We employed a tundish cover ladle with a twin-wire feeding system for nodularization. This modern method offers superior consistency and environmental benefits compared to the traditional sandwich process.
The base iron was melted in a coreless induction furnace. Its target chemistry before treatment is shown in Table 2. Key considerations were a high carbon equivalent for good fluidity and graphitization potential, and very low sulfur and phosphorus levels to minimize the formation of undesirable inclusions and intergranular phases.
| Element | Target (Base Iron) | Final (After Treatment) | Purpose/Effect |
|---|---|---|---|
| Carbon (C) | 3.6 – 3.8 | 3.5 – 3.7 | Graphitizer, fluidity. |
| Silicon (Si) | 1.6 – 1.8 | 2.3 – 2.5 | Graphitizer, strengthens ferrite. |
| Manganese (Mn) | 0.4 – 0.6 | 0.4 – 0.6 | Strengthens pearlite; kept low for toughness. |
| Sulfur (S) | < 0.015 | < 0.015 | Detrimental; consumes Mg. |
| Phosphorus (P) | < 0.05 | < 0.05 | Forms brittle phosphides. |
| Magnesium (Mg) | – | 0.04 – 0.06 | Nodulizing agent. |
| Tin (Sn) | – | 0.05 – 0.07 | Pearlite stabilizer for strength. |
A 700 kg ladle was used. The treatment sequence was as follows: First, a pre-inoculant (0.3%) was added to the empty ladle to prepare the melt for nodularization. The iron was then tapped, leaving enough space in the ladle for the treatment. Immediately after tap, the twin-wire feeder was activated. The cored wire contained an FeSiMg25Re3 alloy (25% Mg, 3% Rare Earths). The amount of wire fed is calculated based on the base iron’s sulfur content and the desired residual magnesium, following a mass balance principle. A simplified version of the calculation is:
$$ m_{wire} = \frac{m_{iron} \times ( [S]_{initial} \times k + [Mg]_{target} )}{\eta \times [Mg]_{wire}} $$
where $m_{wire}$ is wire mass, $m_{iron}$ is iron mass, $[S]_{initial}$ is initial sulfur, $k$ is a factor for Mg consumed by S (typically ~4.5), $[Mg]_{target}$ is target residual Mg (e.g., 0.05%), $\eta$ is Mg recovery efficiency (~40-50% for wire feeding), and $[Mg]_{wire}$ is Mg content in the wire. For our process, the total wire addition was equivalent to 0.7% of the iron weight. The wires inject the alloy deep into the molten iron, ensuring high and consistent recovery. Post-inoculation was performed with 0.2% of a foundry-grade 75% FeSi alloy during transfer to the pouring ladle to enhance graphitization and prevent chilling.
The pouring temperature was strictly maintained between 1430-1450°C. A temperature above this range was corrected by adding clean steel punchings to the ladle. The pour was performed swiftly and steadily, followed by a “topping up” of the risers to ensure they were full. The vacuum was maintained for a full hour after the pour to solidify the casting under pressure and prevent mold collapse or distortion.
Results and Analysis
The castings were shaken out after the required hold time. Visual inspection showed clean surfaces with excellent dimensional accuracy to the pattern. The risers were removed, and their necks revealed sound, dense metal with no visible shrinkage cavities. The castings were shot-blasted, revealing a smooth surface finish.
Test coupons attached to the casting were used for mechanical and metallurgical evaluation. The results, summarized in Table 3, fully met the QT600-3 specification. The microstructure (Fig. 9 in the original text) exhibited well-formed, uniformly distributed graphite nodules (Size 7, Grade 2) in a matrix of approximately 65% pearlite, with the balance being ferrite. This structure is ideal for achieving the required combination of strength, hardness, and pressure tightness. The presence of tin effectively suppressed excessive ferrite formation in these moderately thick sections.
| Property | Result | Specification (QT600-3) |
|---|---|---|
| Tensile Strength (MPa) | 641 | > 600 |
| Yield Strength (MPa) | 420 (Estimated) | > 370 |
| Elongation (%) | 3.5 | > 3 |
| Nodularity Grade | 2 | 1-3 typically acceptable |
| Graphite Size | 7 | – |
| Pearlite Content (%) | ~65 | – |
| Carbides (%) | < 1 | As low as possible |
The ultimate validation came from the machining and pressure testing of the housing. The machined surfaces were free from shrinkage, gas holes, or sand inclusions. The finished component was assembled into the vacuum pump system and subjected to operational pressure tests. It performed flawlessly, with no leakage, confirming the success of the process. The superior erosion resistance of nodular cast iron compared to the original welded steel has led to a dramatically extended service life.
Conclusions and Technical Insights
The successful production of this thick-section vacuum pump housing validates a comprehensive engineering approach to lost foam casting of nodular cast iron. Several key conclusions can be drawn:
1. The lost foam process is exceptionally viable for producing complex, near-net-shape heavy-section castings like this pump housing in nodular cast iron. The combination of CNC foam cutting for pattern segmentation and manual assembly allows for high dimensional precision and is well-suited for low-to-medium volume production runs where hard tooling would be cost-prohibitive.
2. The metallurgical control offered by the tundish cover ladle with twin-wire feeding was instrumental. This method provides precise, reproducible, and efficient magnesium treatment, which is the cornerstone of achieving consistent nodular graphite formation. The high recovery rates and clean operation directly contribute to the mechanical integrity and pressure tightness of the final nodular cast iron casting. The treatment parameters can be summarized by the efficiency equation:
$$ \eta_{Mg} = \frac{[Mg]_{final} \times m_{iron}}{[Mg]_{wire} \times m_{wire}} \times 100\% $$
Our process consistently achieved $ \eta_{Mg} $ between 40-50%.
3. The casting method—a carefully designed top-gating system combined with strategically placed risers—was proven effective through simulation and physical results. The design facilitated calm, sequential filling and established strong thermal gradients for directional solidification. In nodular cast iron, this is coupled with the graphitization expansion that occurs during the last stages of freezing. A simplified model for feeding demand ($V_{feed}$) in a risered system accounts for both liquid contraction ($\beta_l$) and expansion ($\varepsilon_{gr}$):
$$ V_{feed} = V_{casting} \times (\beta_l – \varepsilon_{gr}) $$
where $\beta_l$ is the liquid contraction (∼4% for iron) and $\varepsilon_{gr}$ is the graphite expansion (∼2-4%). Our riser system was designed to provide more than the net $V_{feed}$, resulting in a sound, leak-free casting.
| Factor | Lost Foam (This Study) | Traditional Green Sand |
|---|---|---|
| Pattern/Core Making | Foam assembly; no cores. | Wood/metal pattern; complex core assembly. |
| Dimensional Accuracy | Very High (no parting line shift). | Good, but subject to mold/core shift. |
| Surface Finish | Excellent. | Good. |
| Feeding Design Flexibility | Risers easily placed; chills difficult. | Both risers and chills are applicable. |
| Suitability for Complex Geometry | Excellent. | Good, but limited by core-making. |
In summary, this project demonstrates that a holistic integration of advanced pattern engineering, robust process simulation, and state-of-the-art metallurgical treatment is essential for reliably manufacturing demanding, pressure-tight components in nodular cast iron using the lost foam process. The principles established here—controlled filling, effective feeding to manage solidification, and precise nodularization—form a replicable framework for similar challenging castings.