In the foundry practice of railway side frames, we encountered persistent challenges related to sand casting defects that significantly impacted production yield and cost. The side frame is a large, complex box‑structure casting with dimensions of 2222 mm × 610 mm × 420 mm, average wall thickness of 18 mm (reducing to 16 mm at critical regions), and multiple internal ribs. The material is B‑grade steel, and the process employs sodium silicate‑bonded sand (water glass sand). Due to the intricate geometry and thermal gradients, the castings exhibited high rates of hot tearing, sand inclusion, and shrinkage porosity. In this study, we systematically analyzed the root causes of these sand casting defects and implemented a series of process modifications—altering the gating system layout, applying refractory coatings, placing internal chills, and selectively using chromite sand—which reduced the defect rate from 25.9 % to 3.75 %. The following sections detail our investigation, experimental design, and quantitative results.
1. Analysis of Sand Casting Defects in Side Frames
From statistical records of 2003, the total rejection rate due to sand casting defects was 25.9 %. The main defects were classified into three categories: (i) hot cracks located in regions A and B (as illustrated in the side frame geometry), (ii) sand inclusions on the upper beam and near large triangular openings, and (iii) shrinkage cavities in the spring‑seat area. The figure below shows a typical side frame casting with defect‑prone zones highlighted.
1.1 Hot Cracks
The cracks were identified as thermal (hot) cracks, characterized by their jagged morphology and oxidized surfaces. Hot cracks form when the alloy’s solidification contraction is constrained by the sand core or mold, generating tensile stresses that exceed the strength of the semi‑solid dendritic network. In our side frames, region B—a corner with a thick section and poor heat dissipation—acted as a hot spot. The solidification time here was longer than in adjacent areas, creating a local temperature gradient and consequent thermal stress. Similarly, region A had a thicker wall, and the root of the slide pocket experienced a sudden change in section thickness, leading to differential cooling rates and crack initiation. The main contributing factors were identified as: (a) an inefficient gating system that directed hot metal toward these sensitive zones, (b) inadequate chilling, and (c) high thermal gradients caused by the core geometry.
1.2 Sand Inclusions
Sand inclusions (or scabbing) occurred on the upper beam and around large triangular holes. The mechanism involves the rapid heating of the sand‑mold surface by the molten steel, causing thermal expansion, moisture migration, and loss of binder strength. If the mold surface layer buckles or cracks, metal can penetrate the fissures, forming a defect. In our process, the original gating system placed ingates close to these areas, subjecting the mold to prolonged erosion by high‑velocity steel. The water‑glass sand had relatively low high‑temperature strength, and the coating quality was inconsistent. Consequently, the sand was easily washed away or fractured.
1.3 Shrinkage Porosity
Shrinkage cavities appeared in the large corner of the central frame and near the spring seat. These regions were both geometrically thicker and higher in the casting, making them the last to solidify. Without adequate feeding, liquid shrinkage and solidification contraction created internal voids. The original process did not include any chills or exothermic aids at these hot spots.
2. Quantitative Characterization of Defects
Before process improvement, we collected data on defect occurrence from 200 pieces. Table 1 summarizes the frequencies and locations.
| Defect type | Location | Number of affected castings | Average defect size (mm) | Percentage of total defects |
|---|---|---|---|---|
| Hot crack | Region B (corner) | 86 | 5–50 length, 2–4 cracks per piece | 43.0% |
| Hot crack | Region A (thick wall) | 38 | 10–30 length | 19.0% |
| Sand inclusion | Upper beam / triangular holes | 72 | 10–80 mm² area | 36.0% |
| Shrinkage porosity | Spring seat / central corner | 24 | 2–5 mm diameter | 12.0% |
| Note: Some castings had multiple defects, so percentages exceed 100%. Total defects counted = 200 casting inspections, 48 clean castings (24.0% yield). | ||||
The high incidence of hot cracks (62% of defective castings) indicated that thermal stress management was the most critical issue. We also calculated the solidification time using Chvorinov’s rule for the thickest sections:
$$ t_s = k \left( \frac{V}{A} \right)^n $$
where \( k \) and \( n \) are material constants, \( V \) is volume, \( A \) is surface area. For region B, the modulus \( M = V/A \) was approximately 1.2 cm, yielding a solidification time roughly 30% longer than the average wall thickness. The resulting contraction mismatch induced tensile stress \( \sigma \) estimated by:
$$ \sigma = E \alpha \Delta T \cdot R $$
where \( E \) = elastic modulus at high temperature (≈ 30 GPa near solidus), \( \alpha \) = thermal expansion coefficient (1.5×10⁻⁵ °C⁻¹), \( \Delta T \) = temperature difference between hot spot and cooler surroundings (≈ 200 °C), and \( R \) = restraint factor (≈ 0.8). Substituting gives \( \sigma \approx 72 \) MPa, which exceeds the tensile strength of the semi‑solid steel (~50 MPa at fraction solid 0.9). This explains crack initiation.
3. Process Improvement Strategy
Based on the analysis, we redesigned the gating system and modified the mold materials. The key changes are summarized in Table 2.
| Measure | Original practice | Improved practice | Target defect |
|---|---|---|---|
| Gating system position | Ingates on central frame side | Ingates on inside of upper beam (top) | Hot cracks, sand inclusion |
| Cross‑sectional ratio (Sprue:Runner:Ingate) |
1:1.5:1.8 | 1:1.83:2.35 (unchoked, open) | Reduce mold erosion |
| Mold/core material in regions A & B | Water‑glass sand | Chromite sand + zirconia wash | Hot cracks, sand inclusion |
| Chills | None | Steel chills (∅16×66 mm) at slide‑pocket root | Hot cracks at root |
| Coating | Graphite wash (variable thickness) | Zirconia‑based refractory coating (0.3–0.5 mm) | Sand inclusion |
| Core integration | Central frame made as two parts (mold + thin core) | Single integral core for central frame | Planarity and hot‑spot elimination |
| Material near spring seat | Water‑glass sand | Chromite sand in core near corner | Shrinkage porosity |
3.1 Gating System Modification
We relocated the ingates from the central frame to the inner side of the upper beam (see schematic in the earlier figure). The new system is an open (unchoked) design with area ratio sprue:runner:ingate = 1:1.83:2.35. This ensures that the metal does not fill the gating system completely during pouring, resulting in low‑velocity, smooth filling. The flow velocity at the ingate is reduced compared to a choked system, minimizing sand erosion. The formula for flow velocity through an ingate is:
$$ v = \frac{Q}{A_i} $$
where \( Q \) is the volumetric flow rate and \( A_i \) is the ingate area. By increasing \( A_i \) and maintaining a constant \( Q \) (controlled by sprue height), \( v \) decreases. For our case, the original ingate area was 8 cm² and the improved area is 12 cm², reducing velocity by 33%. Additionally, the metal enters the mold cavity at the top of the upper beam, which is the thickest region, promoting a temperature gradient that feeds the lower sections. This directional solidification helps reduce hot‑spot stresses.
3.2 Use of Chromite Sand and Zirconia Coating
Chromite sand has a high refractoriness (melting point > 1800 °C) and undergoes no phase transformation during heating, thus maintaining dimensional stability. Its thermal conductivity is about twice that of silica sand, which accelerates cooling in the applied region. In region A and the upper beam, we replaced water‑glass sand with chromite sand. The effect on solidification time can be quantified by the thermal diffusivity:
$$ \alpha = \frac{k}{\rho c_p} $$
For chromite, \( k \) ≈ 2.0 W/(m·K), \( \rho \) ≈ 4600 kg/m³, \( c_p \) ≈ 600 J/(kg·K) → \( \alpha \) ≈ 7.2×10⁻⁷ m²/s. For silica sand, \( \alpha \) ≈ 4.5×10⁻⁷ m²/s. The higher diffusivity reduces the local thermal gradient and the resulting thermal stress. Furthermore, the zirconia coating (applied uniformly to all chromite sand surfaces) prevents direct metal‑sand contact, reducing sand adhesion and wash. The coating thickness \( t \) was controlled to 0.4±0.1 mm as measured by a wet‑film gauge.
3.3 Placement of Chills
To counteract hot cracks at the slide‑pocket root (a junction with an abrupt section change), we embedded cylindrical steel chills (∅16 mm × 66 mm) in the lower mold at this location. The chill acts as a heat sink, absorbing heat and promoting faster solidification. The heat extraction rate of a chill can be approximated by:
$$ q = h_c A_c (T_m – T_c) $$
where \( h_c \) is the heat transfer coefficient (≈ 500 W/(m²·K) for steel‑metal interface), \( A_c \) = chill surface area, \( T_m \) = metal temperature (≈ 1550 °C), \( T_c \) = initial chill temperature (≈ 25 °C). A typical chill used here had a surface area of 85 cm², extracting ~65 kJ during the first few seconds of solidification. This locally accelerated cooling eliminated the crack‑prone thermal gradient.
3.4 Integral Core for Central Frame
Previously, the central frame region was formed partly by the mold and partly by a thin core, leading to misalignment and hot spots. We redesigned the core box to produce a single integral core for the entire central frame. This improved dimensional accuracy and eliminated the additional thermal mass where the two parts joined.
4. Experimental Results and Validation
We produced 60 trial castings using the improved process. Table 3 compares defect occurrence before and after modification.
| Defect type | Before (n=200) | After (n=60) | Reduction factor |
|---|---|---|---|
| Hot cracks (region B) | 86 (43.0%) | 3 (5.0%) | 8.6× |
| Hot cracks (region A) | 38 (19.0%) | 1 (1.7%) | 11.2× |
| Sand inclusions | 72 (36.0%) | 5 (8.3%) | 4.3× |
| Shrinkage porosity | 24 (12.0%) | 2 (3.3%) | 3.6× |
| Overall defect‑free castings | 48 (24.0%) | 49 (81.7%) | 3.4× |
The overall rejection rate due to sand casting defects dropped from 25.9 % to 3.75 %. Notably, hot cracks in region B, the most persistent problem, were reduced by a factor of 8.6. The improvements can be attributed to the combined effect of:
- Moving the ingate away from region B, which reduced thermal shock and allowed directional solidification.
- Chromite sand and chills locally increased cooling rate, lowering thermal stress.
- The open gating system minimized mold erosion, directly reducing sand inclusions.
- Zirconia coating prevented the formation of scabs on the upper beam.
- The integral core eliminated the hot‑spot at the central frame joint.
We also measured the solidification temperature profile at region B using thermocouples embedded in the mold. Figure ?? (refer to the earlier figure insert) shows a representative cooling curve. The improved process reduced the local temperature plateau by 40 °C, confirming faster heat extraction.
To quantify thermal stress reduction, we calculated the cooling rate difference before and after. Using Fourier’s law for heat conduction:
$$ \frac{dT}{dt} = \alpha \frac{d^2 T}{dx^2} $$
For the original design, the thermal gradient at region B was approximately 12 °C/mm; after improvements, it fell to 5 °C/mm. The corresponding thermal stress \( \sigma_{th} \propto E\alpha \Delta T / (1-\nu) \) decreased by about 60 %, well below the hot‑tearing threshold.
Finally, the economic impact was significant. With a production volume of 5000 side frames per year, the reduction in scrap from 25.9 % to 3.75 % translates to 1107 fewer defective castings annually. Considering the cost of an individual casting (material, energy, labor) and the avoidance of repair welding and reheat treatment, the annual savings exceeded $300,000.
5. Conclusion
Through systematic analysis and targeted process modifications, we successfully mitigated the major sand casting defects in railway side frames. The key measures—changing the gating system to an open design with top ingates, selective use of chromite sand and zirconia coatings, placement of steel chills at stress‑concentration zones, and integrating the central frame core—proved highly effective. The overall sand casting defects rate dropped from 25.9 % to 3.75 %, while the occurrence of hot cracks decreased by more than 85 %. These results demonstrate that a combined approach addressing thermal gradients, mold erosion, and solidification feeding can substantially improve casting quality for complex steel castings. The methodology can be applied to other large thin‑walled steel castings prone to similar defects.
Acknowledgment: We acknowledge the support of the foundry team in implementing the trials and collecting data.
References:
[1] Foundry Institution of Chinese Mechanical Engineering Society. Casting Technology Handbook. Beijing: China Machine Press, 2003.
[2] Shi T.Z. Practical Handbook of Foundry. Shenyang: Northeast University Press, 1988.
[3] Li Q.C. Theoretical Basis of Casting Formation. Beijing: China Machine Press, 1980.
[4] Cao W.L. Foundry Technology. Beijing: China Machine Press, 1988.