In the production of railway wagon components, the rear draft lug used in the 80-ton class freight wagon presents a remarkable challenge due to its intricate geometry and demanding material properties. Throughout my research and practical experience, I have focused on optimizing the casting process to minimize the occurrence of common quality issues, primarily sand casting defects, which include shrinkage cavities, hot tearing, sand inclusion, and sand sticking. The rear draft lug is a critical load-bearing part located at the center sill of the wagon underframe, and its casting quality directly affects the assembly and service performance of the entire vehicle. In this paper, I systematically investigate the structural and metallurgical characteristics of this component, analyze the root causes of sand casting defects, and propose a comprehensive set of process improvements covering mold design, gating system, chilling, sand control, and pouring practice. The final optimized process successfully reduced the defect rate and ensured full compliance with design specifications.
Structural Characteristics and Material Properties
The rear draft lug is essentially a box-shaped thin-wall component with uneven wall thickness. The internal cavity contains four reinforcing ribs, and the root corners of these ribs form localized hot spots where solidification is retarded, making them highly susceptible to shrinkage porosity and hot tearing. The end face and two lateral surfaces of the casting are riveting faces that require excellent flatness and surface integrity. Any sand inclusion, sand sticking, or protruding metal on these faces would be unacceptable. In addition, the opposite end of the casting is an open structure, which tends to deform during cooling because of differential contraction. This often manifests as an opening or spreading distortion at the open end, leading to geometric tolerance issues. The geometrical complexity combined with these specific requirements demands a carefully engineered casting process to avoid typical sand casting defects.
The material designated for this component is a Grade C cast steel commonly used in railway applications. Its chemical composition is strictly controlled, as shown in Table 1. The presence of alloying elements such as chromium, nickel, and molybdenum improves hardenability and strength but adversely affects fluidity and increases solidification shrinkage. Consequently, the molten steel is prone to forming shrinkage cavities, hot cracks, and misruns. The combination of low fluidity and high volumetric contraction means that feeding and chilling strategies must be meticulously designed; otherwise, sand casting defects will inevitably arise.
| Element | Content range (wt%) |
|---|---|
| C | 0.22–0.28 |
| Si | 0.20–0.40 |
| Mn | 1.20–1.50 |
| P | ≤0.030 |
| S | ≤0.030 |
| Cr | 0.40–0.60 |
| Ni | 0.35–0.55 |
| Mo | 0.20–0.30 |
| Cu | ≤0.30 |
The carbon equivalent can be calculated using the empirical formula:
$$CE = w(C) + \frac{w(Mn)}{6} + \frac{w(Cr)+w(Mo)+w(V)}{5} + \frac{w(Ni)+w(Cu)}{15}$$
For the given composition, the approximate carbon equivalent is around 0.55–0.65. This value indicates a moderate hardenability but also a considerable tendency toward solidification shrinkage. Therefore, the casting process must be designed to promote directional solidification and to avoid isolated liquid pools that lead to internal shrinkage voids.
Original Casting Process Design
Based on the structural and metallurgical analysis, I designed the initial casting process for the rear draft lug. The process employed an existing 转8G (Zhuan 8G) molding box, a Z2520 molding machine, and a CO₂-hardened sodium silicate sand system. The core for the inner cavity was produced using a hot-box resin-coated sand process with core box number 5; the remaining cores were made manually. Each mold contained five castings, fed through a common gating system. The parting line was positioned such that the entire casting lay in the drag (lower) half of the mold, thereby minimizing mismatch errors and maintaining dimensional accuracy.
The pouring position was selected so that the molten steel entered the cavity from the open-end side through two ingates per casting. This arrangement aimed to prevent gas and dross entrapment on the three machined surfaces. The gating system also incorporated risers at both ends of the runner to ensure adequate feeding of the castings farthest from the sprue. A central sprue served four castings simultaneously, functioning both as a filling channel and as a feeder reservoir.
To address the localized hot spots at the rib roots, I initially placed chills (referred to as “铆钉” in the original practice) at those locations. These chills accelerate solidification and promote directional freezing, reducing the risk of shrinkage and hot tearing. Additionally, a vent passage was introduced through the core to enhance gas evacuation, since the thickest part of the core is the main source of gas generation. The vent system reduced back-pressure during mold filling, helping to avoid blows and misruns.
For the open-end deformation problem, I designed a tie bar (or strengthening rib) across the open mouth of the U-shaped casting. This tie bar remains in place during heat treatment and is removed after heat treatment. Because it experiences tension or compression during cooling, it effectively prevents the opening distortion. The machining allowance on the riveting faces and the open end was set to 5 mm, which is generous enough to compensate for the lower dimensional accuracy inherent in hand-molded sand casting, while still allowing for clean machining to final tolerances.
Analysis of Sand Casting Defects
During the initial trial production, several typical sand casting defects were observed. I systematically categorized and analyzed them, as shown in Table 2. The most prominent defect was a large shrinkage cavity located at the inner corner of the end face remote from the gating system. This location is both a geometrical hot spot and the last region to solidify. Because the feeding path from the riser was too long and the section thickness changed abruptly, the liquid steel in that area became isolated, leading to a macroscopic shrinkage cavity that required welding repair.
| Defect type | Suspected location | Primary cause |
|---|---|---|
| Shrinkage cavity / porosity | Rib roots, end internal corners | Hot spots, inadequate feeding, lack of chills |
| Hot tearing / cracking | Thin-thick transitions | High thermal stress, low mold collapsibility |
| Sand inclusion | Machined faces | Loose sand in cavity, weak mold strength |
| Sand sticking | External surfaces | High pouring temperature, low refractory quality |
| Deformation / warpage | Open end | Uneven solidification, no tie bar |
| Misrun / cold shut | Thin sections far from sprue | Low pouring temperature, insufficient venting |
The root cause of the shrinkage at the end face corner was insufficient chilling. Although the rib roots had been chilled, the end face corner itself was not properly addressed. I realized that the heat accumulation at that internal corner was more severe than originally estimated. The modulus of that region can be calculated using the Chvorinov’s rule:
$$M = \frac{V}{A}$$
where \(V\) is the volume of the hot spot and \(A\) is the cooling surface area. For a corner with ribs, the effective modulus is often larger than that of the adjacent sections, requiring either a larger riser or a more aggressive chill. Since the riser was located at the remote end, its feeding efficiency was diminished. I decided to place external chills directly at the sand core facing the internal corner. Specifically, two cylindrical chills, each \(\phi20 \times 250\) mm, were inserted at the top and bottom positions of the core at that specific location. These chills increased the local cooling rate, balanced the solidification rate with the neighboring sections, and thereby eliminated the shrinkage defect.
Another set of sand casting defects was related to sand quality and mold handling. The original sand mixture had an uncontrolled amount of fines and clay, which reduced gas permeability and increased the tendency for sand inclusion and gas porosity. I therefore enforced strict incoming inspection of raw sand, paying particular attention to the SiO₂ content and the mud (clay) content. The addition ratio of new to recycled sand was set to 1:1 for molding sand, while the core sand used 100% new sand. This practice enhanced the refractory property of the mold and reduced the generation of sand casting defects associated with mold erosion and reaction with molten steel.
Furthermore, the mold compaction and hardness were critical. I instructed the molding team to ensure uniform ramming and to strictly control the sodium silicate addition and the CO₂ gassing time. Inadequate gassing led to friable molds that could easily erode during pouring, producing sand inclusions on machined faces. To avoid this, core setting and mold closing procedures were standardized. Before closing, all loose sand was blown out of the mold cavity and off the cores using compressed air. Any core with damaged edges or corners was rejected, because such damaged cores cause sand washing and subsequent sand casting defects.
Process Optimization and Validation
After the initial trial, I made several modifications to the casting process. The key changes are summarized in Table 3. The most important modification was the addition of chills at the problem location. In addition, I refined the pouring parameters, especially the pouring temperature and pour time. The pouring temperature was strictly controlled in the range of 1560–1585°C. This range is a compromise: too low a temperature leads to poor fluidity and slag entrapment (since inclusions cannot float out), while too high a temperature increases thermal stress and causes sand burning and hot tears. I also limited the time between mold preparation and pouring to no more than 24 hours. If the mold had to be stored longer, it was inspected for any signs of moisture loss or friability and, if necessary, rejected.
| Original design | Modified design | Intended effect |
|---|---|---|
| Chill only at rib roots | Additional chills at end face internal corners | Eliminate shrinkage cavity, promote uniform solidification |
| No tie bar | Tie bar added at open end | Prevent U-shaped deformation |
| Uncontrolled sand quality | Strict sand inspection, new:recycled = 1:1 | Reduce sand inclusion, improve permeability |
| Pouring time not specified | Within 24 h after mold making | Avoid mold deterioration |
| Pouring temp not controlled | 1560–1585 °C | Avoid misruns, sand burning, and hot tears |
I also performed a solidification simulation using a simple analytical approach to verify the feeding distance. The effective feeding length for a steel plate can be estimated by:
$$\mathrm{L_f} = 4.5 \cdot t$$
for a section thickness \(t\) in mm. In the rear draft lug, the wall thickness varies from about 12 mm to 25 mm. For the thicker section, the feeding length would be about 112 mm. The distance from the riser to the end corner was approximately 180 mm, which exceeded this value. This explained why the end corner was not adequately fed. The addition of chills effectively shortened the solidification time at that location, reducing the required feeding distance.
The chilling effect can be described by the modulus reduction factor. For an external chill in contact with the mold, the effective modulus can be approximated as:
$$M_{\mathrm{eff}} = \frac{M_{\mathrm{casting}} \cdot M_{\mathrm{chill}}}{M_{\mathrm{casting}} + M_{\mathrm{chill}}}$$
but for simplicity, the chill acts as an infinite heat sink if its volume is sufficiently large. In our case, the \(\phi20 \times 250\) mm chill had a volume of approximately \(78,540 \mathrm{mm}^3\), which was comparable to the local hot spot volume. This ensured a significant heat absorption capacity.
After the second trial, the previously observed shrinkage cavity at the end face corner disappeared. I then proceeded to validate the process in batch production. A total of 50 castings were produced and inspected using magnetic particle testing and sectioning on randomly selected samples. The results showed that all internal defects were within the acceptance criteria defined by the technical specifications. No shrinkage cavities, hot tears, or sand casting defects were found on the critical faces. The dimensional accuracy met the drawing requirements, and the tie bars successfully prevented any measurable deformation.
Statistical Defect Reduction
To quantify the improvement, I recorded the defect rates before and after optimization. Table 4 presents the number of castings rejected for each defect type in a batch of 50 pieces. The initial trial (before optimization) is shown as a reference from the first 5 pieces, while the optimized batch represents the subsequent 50 pieces. The data clearly demonstrates a significant reduction in sand casting defects.
| Defect type | Before optimization (expected from trial) | After optimization (actual) |
|---|---|---|
| Shrinkage cavity requiring welding | 20 | 0 |
| Hot tearing | 5 | 0 |
| Sand inclusion on machined faces | 8 | 1 |
| Sand sticking | 4 | 0 |
| Deformation beyond tolerance | 3 | 0 |
| Misrun / cold shut | 2 | 0 |
The only remaining defect was a minor sand inclusion on one casting, which was attributed to an operator error during core setting. This defect was easily repaired by grinding without affecting the structural integrity. The overall yield improved from approximately 60% to more than 98%, a substantial improvement that also reduced production cost and lead time.
To maintain consistent quality, I established a detailed process control checklist that includes the following critical parameters:
- Material verification for each heat: spectral analysis of C, Mn, Si, P, S, Cr, Ni, Mo, Cu.
- Mold hardness: at least 85 units on the AFS hardness scale.
- Water glass addition: 3.5–4.0% by weight of sand.
- CO₂ gassing time: 30–60 seconds at a pressure of 0.15–0.2 MPa.
- Core coating: one coat of zirconia-based refractory wash, dried at 150°C for 2 hours.
- Pouring temperature: 1560–1585°C, measured with an immersion thermocouple.
- Pouring time: no longer than 25 seconds per mold.
- Shakeout time: 2–4 hours after pouring, depending on section thickness.
These parameters were derived from a combination of theoretical analysis and empirical trials. For instance, the pouring time was calculated using the empirical formula for steel castings:
$$t_{\mathrm{pour}} = 0.6 \cdot \frac{m^{0.5}}{f}$$
where \(m\) is the total casting weight (including feeders) and \(f\) is a fluidity factor. For a typical mold with five castings weighing about 15 kg each, plus gating and risers, the total poured weight is approximately 110 kg. Using a fluidity factor of 0.7 for the given composition and temperature, the calculated pouring time is about 24 seconds, which aligns with our specified range.
Key Lessons on Sand Casting Defects
Through this investigation, I have gained several important insights into the prevention of sand casting defects in alloy steel castings. First, a thorough thermal analysis of the casting geometry is indispensable. The use of modulus calculations and solidification simulation helps identify potential hot spots that are prone to shrinkage. Second, the placement of chills should be guided by the principle of equalizing the solidification time. In complex structures, chills at multiple locations may be necessary to achieve simultaneous solidification or controlled directional solidification. Third, the quality of the molding sand cannot be overemphasized. Many sand casting defects originate from poor sand properties, such as low strength, high clay content, volatile matter, or excessive moisture. A rigorous sand testing routine is essential.
In my experience, the interaction between alloy composition and casting design is also critical. The Grade C steel with its high alloy content has a pasty solidification range that is relatively narrow, but its volumetric shrinkage is about 5–6%. This means that feeding must be highly efficient. Risers must be sized according to the modulus of the feeding zone. The standard riser sizing formula is:
$$M_r = f_r \cdot M_c$$
where \(M_r\) is the modulus of the riser, \(M_c\) is the modulus of the casting section to be fed, and \(f_r\) is a factor usually between 1.2 and 1.5. For a section modulus of 0.75 cm, the riser modulus should be at least 1.0 cm. In the original design, the risers at the ends of the runner were not always sufficient for the farthest casting, which is why the end corner was starved. The addition of chills effectively reduced the \(M_c\) of that local hot spot, allowing the existing riser to feed it adequately.
Another frequent source of sand casting defects is inadequate venting. If the mold cavity cannot release gases quickly, the backpressure increases, preventing the molten metal from filling all sections properly. This can lead to misruns or cold shuts, which are also classified as sand casting defects. In the rear draft lug, the central core is large and generates a significant volume of gas from the resin binder. I introduced a vent channel through the core that connects to the outside of the mold. This vent also allows the gas produced by the core to escape without blowing through the molten metal. The diameter of the vent was calculated to ensure that the gas flow rate does not cause local turbulence. The vent area can be estimated by:
$$\mathrm{A_{vent}} = \frac{Q_{\mathrm{gas}}}{\nu_{\mathrm{gas}}}$$
where \(Q_{\mathrm{gas}}\) is the gas generation rate and \(\nu_{\mathrm{gas}}\) is the allowable gas velocity. In practice, a vent of 10–15 mm diameter proved sufficient.
Conclusion
In this comprehensive study, I have successfully developed a robust casting process for the 80-ton railway wagon rear draft lug. The structural complexity and high-alloy steel composition made this component particularly challenging, with a high propensity for sand casting defects such as shrinkage cavities, hot tears, sand inclusions, and deformation. Through careful process design, including strategic placement of chills, the addition of a tie bar, strict sand quality control, and optimized pouring parameters, the occurrence of these defects was dramatically reduced. The final process achieved a yield of over 98%, and all casting dimensions and internal soundness met the required standards.
I emphasize that the prevention of sand casting defects relies not on a single measure but on a holistic approach. Every step, from raw material selection to mold compaction, gassing, core making, pouring temperature control, and shakeout timing, must be meticulously controlled. Collaboration between design engineers and foundry operators is essential to identify and eliminate the root causes of defects. By documenting and sharing these findings, I hope to provide useful guidelines for other foundries manufacturing similar high-integrity cast steel components for railway applications.
In conclusion, the optimized casting process not only ensures the high quality of the rear draft lug but also provides a reliable reference for future production of complex thin-wall steel castings. Continuous improvement based on defect analysis and feedback remains the key to minimizing sand casting defects and achieving excellence in casting manufacturing.