In the field of metal casting, the production of large-scale thin-walled components, especially those made from nodular cast iron, presents significant challenges due to the material’s unique solidification characteristics. Nodular cast iron, also known as ductile iron, is renowned for its high strength, ductility, and toughness, which stem from the spheroidal graphite morphology achieved through magnesium or cerium treatment. However, the casting process for complex geometries, such as box covers for power locomotives, often encounters defects like shrinkage cavities and porosity, which can compromise structural integrity and performance. This article details our first-person experience in improving the casting process for a nodular iron box cover using Computer-Aided Engineering (CAE) simulation technology. The initial production process led to shrinkage defects in thick sections, which were systematically analyzed and resolved through CAE-assisted design modifications, ultimately yielding合格 castings that met stringent ultrasonic testing standards.
The box cover in question is a critical component for a power locomotive, with a diameter of approximately 1500 mm, a height of 700 mm, and an average wall thickness of 20 mm. The casting weight is 680 kg, and the material specification is QT400-15, a grade of nodular cast iron with a tensile strength of 400 MPa and 15% elongation. The component must undergo full ultrasonic non-destructive testing (UT) to ensure the absence of shrinkage cavities and porosity, with overall UT ratings not exceeding CC2 level according to industry standards. The geometry features several thick sections, particularly at the junctions where the cover connects to other parts, which act as thermal hubs during solidification. These areas are prone to shrinkage defects due to the extended solidification time and the complex interplay between liquid metal feeding and graphite expansion in nodular cast iron.
Originally, the casting process employed a furan resin sand molding and core-making system, with a resin addition of 1.01% relative to sand weight and a catalyst at 44.44% of resin weight. The mold properties included a tensile strength of 0.20–0.55 MPa, compressive strength of 0.8–1.7 MPa, permeability of 350–1000, and gas evolution below 19 mL/g. A graphite coating of 0.2–0.3 mm thickness was applied. The pattern layout was one casting per mold box, with a horizontal parting plane. The gating system was semi-closed, with a ratio of sprue cross-sectional area to runner cross-sectional area to ingate cross-sectional area set at 1.6:2:1. Specifically, the sprue area was 5024 mm², the runner area was 6358 mm², and the ingate area was 3200 mm². Side risers were used, along with two ceramic filters. Pouring temperature ranged from 1360 to 1390°C, with a pouring time of about 35 seconds.
Melting was conducted in a 6-ton medium-frequency induction furnace, with a charge composition of 55% pig iron, 30% scrap steel, and 10–20% returns. The target chemical composition for the nodular cast iron was: carbon (C) 3.6–4.0%, silicon (Si) 2.2–2.5%, manganese (Mn) <0.35%, phosphorus (P) <0.05%, sulfur (S) <0.015%, and magnesium (Mg) 0.03–0.05%. Nodularization was achieved via the sandwich method using a rare-earth magnesium alloy added at 1.0%, followed by inoculation with YFY-4 inoculant (5–15 mm grain size) at 0.8% and secondary stream inoculation with YFY-2 inoculant (0.2–0.8 mm grain size) at 0.15%.
Despite these meticulous parameters, initial trial castings exhibited shrinkage cavities in thick sections, as detected by ultrasonic testing. The casting was divided into numbered zones for systematic evaluation, and defects were recorded in areas corresponding to thick junctions and adjacent walls. For instance, zones #52, #55, #104, #172, #175, and #176 showed shrinkage levels exceeding CC2, reaching up to CC5 in some cases. This prompted a detailed investigation using CAE simulation software to model the solidification process and identify the root cause.
CAE technology has revolutionized foundry practices by enabling virtual prototyping of casting processes, predicting defects, and optimizing designs before physical production. For nodular cast iron, the simulation must account for the graphite expansion during eutectic solidification, which can compensate for shrinkage but also lead to secondary contraction if not properly managed. We employed a commercial CAE tool to simulate the original process, incorporating thermophysical properties of nodular cast iron, mold materials, and boundary conditions. The simulation output included temperature fields, solidification sequences, and shrinkage porosity predictions.
The CAE results revealed that the thick junctions solidified last, creating thermal hubs where liquid metal remained isolated from feeding sources. In nodular cast iron, the graphite precipitation during eutectic reaction generates an expansion force that can push liquid metal into adjacent regions. However, in our case, the expansion from late-solidifying areas fed metal into the thick box cover junctions, but when these junctions themselves underwent secondary shrinkage, the already solidified graphite could not force the diminishing liquid back, resulting in shrinkage cavities. This phenomenon is captured by the following equation for volume change during solidification:
$$ \Delta V = V_L \cdot \alpha_L \cdot (T_L – T_S) + V_S \cdot \beta_S – V_G \cdot \epsilon_G $$
where \(\Delta V\) is the net volume change, \(V_L\) is the liquid volume, \(\alpha_L\) is the liquid thermal contraction coefficient, \(T_L\) and \(T_S\) are liquidus and solidus temperatures, \(V_S\) is the solid volume, \(\beta_S\) is the solid contraction coefficient, \(V_G\) is the graphite volume, and \(\epsilon_G\) is the graphite expansion coefficient. For nodular cast iron, \(\epsilon_G\) is significant due to spheroidal graphite formation, but its effectiveness depends on timing and proximity to liquid pools.
The simulation also highlighted that areas near thick sections, such as zones #68, #79, and #93, developed shrinkage due to delayed solidification. To quantify this, we analyzed the solidification time using Chvorinov’s rule, modified for nodular cast iron:
$$ t_f = K \cdot \left( \frac{V}{A} \right)^n $$
where \(t_f\) is the local solidification time, \(V\) is the volume of the section, \(A\) is the surface area, \(K\) is a constant dependent on mold material and casting conditions, and \(n\) is an exponent typically around 2. For the thick junctions, the \(V/A\) ratio was high, leading to prolonged \(t_f\) and increased shrinkage risk. The table below summarizes the solidification times and defect levels for key zones:
| Zone Number | Section Thickness (mm) | Calculated V/A Ratio (mm) | Estimated Solidification Time (s) | Shrinkage Level (UT) |
|---|---|---|---|---|
| #52 | 40 | 20.5 | 210 | CC5 |
| #55 | 38 | 19.8 | 205 | CC4 |
| #79 | 35 | 18.2 | 190 | CC5 |
| #93 | 36 | 18.5 | 192 | CC5 |
| #104 | 42 | 21.0 | 215 | CC5 |
| #172 | 41 | 20.8 | 212 | CC5 |
| #175 | 39 | 20.0 | 208 | CC5 |
| #176 | 37 | 19.5 | 200 | CC4 |
Based on this analysis, we devised an optimization strategy centered on accelerating solidification in thick regions to balance the overall cooling and eliminate thermal hubs. The solution involved placing chills at five critical thick locations and at the box cover junctions. Chills are metallic inserts with high thermal conductivity that extract heat rapidly, reducing local solidification time. The design considered chill size, material (typically iron or copper), and placement to avoid over-cooling that could cause cracks. The modified casting layout included chills with dimensions tailored to each hot spot, as shown in the following specification table:
| Chill Location (Zone) | Chill Material | Dimensions (mm) | Contact Area (mm²) | Expected Cooling Rate Increase (%) |
|---|---|---|---|---|
| Junction #52 | Cast Iron | 100 x 50 x 25 | 5000 | 40 |
| Junction #55 | Cast Iron | 90 x 45 x 20 | 4050 | 35 |
| Near #104 | Copper | 80 x 40 x 15 | 3200 | 50 |
| Junction #172 | Cast Iron | 95 x 48 x 22 | 4560 | 38 |
| Junction #175 | Cast Iron | 85 x 42 x 20 | 3570 | 36 |
The effectiveness of chills can be modeled using heat transfer equations. For a chill in contact with molten nodular cast iron, the heat flux \(q\) is given by:
$$ q = h \cdot (T_m – T_c) $$
where \(h\) is the heat transfer coefficient, \(T_m\) is the metal temperature, and \(T_c\) is the chill temperature. Integrating over time, the total heat extracted \(Q\) reduces the local solidification time according to:
$$ Q = \int_0^{t_f} q \, dt = \rho \cdot V \cdot L_f $$
with \(\rho\) as density and \(L_f\) as latent heat of fusion. By increasing \(q\) via chills, \(t_f\) decreases, promoting simultaneous solidification across the casting.
We re-ran the CAE simulation with the chills incorporated, and the results showed a dramatic reduction in predicted shrinkage. The solidification sequence became more uniform, with no isolated liquid pockets in the thick junctions. The modified process was then implemented in production. The chills were positioned in the mold cavity during molding, ensuring good contact with the sand. All other parameters, including melting, nodularization, and pouring, remained unchanged to isolate the effect of the chills.
Post-optimization castings were subjected to ultrasonic testing, and all zones reported shrinkage levels within CC2, meeting the specification. The table below compares defect levels before and after optimization for selected zones:
| Zone Number | Shrinkage Level (Original) | Shrinkage Level (Optimized) | Improvement |
|---|---|---|---|
| #52 | CC5 | CC1 | 4 levels |
| #55 | CC4 | CC1 | 3 levels |
| #79 | CC5 | CC2 | 3 levels |
| #93 | CC5 | CC2 | 3 levels |
| #104 | CC5 | CC1 | 4 levels |
| #172 | CC5 | CC1 | 4 levels |
| #175 | CC5 | CC1 | 4 levels |
| #176 | CC4 | CC1 | 3 levels |
The success of this improvement underscores the value of CAE technology in modern foundries, particularly for complex nodular cast iron castings. By simulating the solidification process, we could pinpoint defect mechanisms and test solutions virtually, saving time and resources compared to trial-and-error methods. The use of chills to balance solidification is a classic technique, but CAE enabled precise placement and sizing for maximum efficacy.
Beyond this specific case, the principles apply broadly to nodular cast iron components with varying section thicknesses. The graphite expansion behavior must be harnessed through controlled cooling, and CAE simulations can model this dynamically. Future work could explore advanced feeding systems, such as pressurized risers or exotic chill materials, to further enhance quality. Additionally, integrating CAE with real-time process monitoring could lead to adaptive control systems that adjust pouring parameters based on simulation feedback.
In conclusion, the casting process for a nodular iron box cover was successfully improved by employing CAE technology to analyze and address shrinkage defects. The original process, while sound in general, failed in thick sections due to unbalanced solidification and secondary contraction. Through simulation, we identified that graphite expansion from late-solidifying areas fed into junctions, but subsequent shrinkage there created cavities. By adding strategically placed chills, we accelerated cooling in these regions, harmonized the solidification timeline, and eliminated shrinkage. The optimized castings passed ultrasonic testing, validating the approach. This experience highlights how CAE-driven design can solve practical foundry problems, ensuring high-quality nodular cast iron components for demanding applications.