Investment casting is used for low-alloy steel railway bogie mounts because it can form thin walls, internal cavities, and local thick bosses with limited machining. The process window is narrow. The mount is a steering-frame component that transmits traction and braking loads; its 410 × 296 × 330 mm envelope contains walls from 8 to 45 mm and R5 transitions between thick and thin sections. These transitions are both hot spots and restraint points. Under a baseline shell condition, magnetic-particle inspection of production mounts showed cracks concentrated at riser junctions and thick-to-thin transitions, with a crack rate above 50%.
By 2020, China’s high-speed rail network reached 37,900 km; annual production of high-speed train sets in 2016–2019 was about 3,000 cars, and domestic demand for rail bogie castings has been estimated at more than 20,000 t per year. The mount is made from C-grade ZG25MnCrNiMo steel under TB/T 2942-1999, with a specified minimum yield strength of 415 MPa, tensile strength of 620 MPa, elongation of 22%, reduction of area of 45%, and Charpy impact energy of 20 J at −18°C. Those are finished-casting requirements; the as-cast test bars discussed below were tested before heat treatment, so their absolute ductility is not directly comparable with the specification.
Casting System and Solidification Path
Three top-poured designs were evaluated. The first used four bottom risers, two top pouring-feeding risers and an enlarged runner. It fed the casting but had low yield and slight porosity at the top riser. The second removed two bottom risers and added a side riser; it produced porosity above the circular hole and a high stress concentration. The final design connected a 60 × 70 × 190 mm side pouring-feeding riser to a 60 × 70 × 160 mm top riser, added two 60 × 70 × 110 mm side risers, and used a top pad to feed the thick upper section. This gave the best combination of feeding and yield.
Filling was simulated with a 25 mm shell, a 500 W/(m²·K) metal–shell interface, air cooling, 1560°C metal, 5.10 kg/s, 950°C shell preheat, and 300°C shell at pour. The 21 s fill was smooth: metal entered the runner, buffered in the side riser, and rose through the cavity without splashing.
Solidification followed the intended bottom-to-top sequence. At 114.2 s, the middle thin wall was about 0.8 solid while regions near the runners and risers were below 0.4; the casting was fully solid at 404.3 s. The last liquid was in the top pouring riser, and the feed path remained open. Shrinkage porosity was therefore confined to the riser system. At 1560°C no internal porosity was predicted. Raising the metal temperature to 1580°C produced 2.76 cm³ of shrinkage, about 0.04% of casting volume, and 1600°C produced 5.32 cm³, about 0.07%.
Thermal Stress and Process Sensitivity
The baseline stress calculation used 1560°C metal, 5.10 kg/s, 950°C shell preheat and 300°C shell at pour. The effective stress was non-uniform. The peak was 827.50 MPa at the riser–casting junction; secondary concentrations appeared in the thin lower wall, which solidified first and then resisted contraction of the thicker sections. The computed peak exceeded the measured tensile strength of the room-temperature-shell as-cast material, 737.72 MPa, which is consistent with the high crack rate observed in production.
Stress evolution had three stages. During solidification, the rising solid fraction increased interlocking and shell restraint, so stress rose quickly. After solidification, the austenite-to-ferrite/carbide transformation produced a volume change that was also constrained, and the stress rate reached a maximum. Once transformation and cooling were complete, the rate declined. For a simple free-contraction estimate, the thermal strain is:
$$ \varepsilon = \alpha (T_p – T_s) $$
where \(T_p\) is the pouring temperature, \(T_s\) the solidus temperature, and \(\alpha\) the linear expansion coefficient. Higher pouring temperature increases the free contraction strain; because the shell and cooler sections restrain that contraction, the local stress rises. The relation is not linear in a real casting because \(\alpha\) and restraint vary with temperature, but it explains the positive trend with metal temperature.
| Parameter | Range tested | Peak effective stress | Observed effect |
|---|---|---|---|
| Shell temperature at pour | 200–700°C | 832.74 → 689.59 MPa | Dominant variable; strong reduction with increasing shell temperature |
| Metal pouring temperature | 1550–1600°C | 826.23 → 842.30 MPa | Modest increase; +16.07 MPa, about 1.94% |
| Pouring rate | 4.70–5.70 kg/s | 830.52 → 822.64 MPa | Weak effect; maximum spread 9.02 MPa and no monotonic trend |
Shell temperature was the dominant variable. Increasing shell temperature at pour from 200 to 700°C lowered the peak from 832.74 to 689.59 MPa, a reduction of 143.15 MPa. Most of the drop occurred between 300 and 600°C, where the peak fell from 821.50 to 708.34 MPa. Metal temperature had a smaller, positive effect: 1550 to 1600°C raised the peak from 826.23 to 842.30 MPa. The trend was not strictly monotonic; the 1580°C case was slightly lower than 1570°C. Pouring rate had little influence: between 4.70 and 5.70 kg/s the peak varied within a 9.02 MPa band.
On this basis, the selected condition was 1560°C metal, 5.5 kg/s, 950°C shell preheat and 700°C shell at pour. The predicted peak was 689.59 MPa, and production trials showed a marked reduction in crack rate relative to the 300°C-shell baseline.
Shell Temperature, Microstructure, and Tensile Properties
To isolate the metallurgical effect of shell temperature, test bars were cast with the mounts under room-temperature shell and 400°C shell. The bars had the same chemistry as the mount: about 0.22% C, 1.36% Mn, 0.46% Cr, 0.43% Ni, 0.27% Mo and 0.30% Si, with S and P near 0.02%. Tensile testing was at room temperature at 4 mm/min, with four bars per condition.
With a room-temperature shell, fast cooling produced blocky ferrite, Widmanstätten ferrite, granular bainite, a small amount of retained austenite, and rounded inclusions. Widmanstätten ferrite forms when austenite decomposes during relatively fast cooling and ferrite grows along austenite grain boundaries into the grain interior; Mn in the steel promotes it. Granular bainite appeared as island-like regions in a ferrite matrix. EDS mapping showed Mn and Mo segregation in these islands, while Cr and Ni were more uniformly distributed.
With a 400°C shell, slower cooling gave more time for carbon diffusion. The coarse Widmanstätten structure disappeared, blocky ferrite was replaced by acicular ferrite, and pearlite formed along with finer granular bainite and retained austenite.
| Constituent | Room-temperature shell (HV0.1) | 400°C shell (HV0.1) |
|---|---|---|
| Blocky ferrite / acicular ferrite | 158.2 | 172.8 |
| Widmanstätten ferrite / pearlite | 262.5 | 287.3 |
| Granular bainite | 346.4 | 320.7 |
| Inclusions | 486.5 | 479.0 |
Microhardness tracked the constituents. Blocky ferrite was the softest at 158.2 HV; acicular ferrite was 172.8 HV. Widmanstätten ferrite was 262.5 HV, and pearlite 287.3 HV. Granular bainite was 346.4 HV in the room-temperature-shell condition and 320.7 HV at 400°C. Inclusions were about 480–487 HV. The hard inclusions were rounded, so their effect on matrix continuity was limited, but they remain local stress raisers.
| Shell condition | Main constituents | Yield strength (MPa) | Tensile strength (MPa) | Elongation | Brinell hardness (HBW) |
|---|---|---|---|---|---|
| Room temperature | Blocky ferrite, Widmanstätten ferrite, granular bainite, retained austenite | 648.34 | 737.72 | <1% | 272.7 |
| 400°C | Acicular ferrite, pearlite, granular bainite, retained austenite | 663.51 | 834.04 | ~2.8% | 263.6 |
At room-temperature shell, tensile strength was 737.72 MPa, yield strength 648.34 MPa and elongation below 1%; Brinell hardness was 272.7 HBW. At 400°C shell, tensile strength rose to 834.04 MPa, an increase of 96.32 MPa or 13.06%; yield strength rose to 663.51 MPa, an increase of 15.17 MPa; elongation reached about 2.8%, and hardness fell to 263.6 HBW. The strength gain is consistent with finer acicular ferrite and more complete transformation; the ductility gain is consistent with removal of coarse Widmanstätten ferrite.
Fracture of the room-temperature-shell bars showed river patterns typical of cleavage, with little plastic deformation. EDS of dark particle regions showed elevated C, O, Mn and N, indicating nonmetallic inclusions. Such inclusions interrupt the matrix and can initiate cleavage cracks. The 400°C condition, with higher elongation, is less prone to this brittle failure mode.
Crack Mechanism and Control Levers
The crack mechanism is a coincidence of mechanical and metallurgical factors. The mechanical factor is the computed stress concentration at riser junctions and section transitions: 827.50 MPa at the baseline, above the 737.72 MPa tensile strength of the room-temperature-shell as-cast material. The metallurgical factor is the low ductility of that microstructure—less than 1% elongation and cleavage fracture. Once the local stress exceeds the local strength, a crack initiates at a riser junction, inclusion, or thin-wall transition and propagates without much plastic blunting.
Raising shell temperature acts on both sides. It reduces thermal gradients and cooling rate, lowering the computed peak to 689.59 MPa at 700°C. It also changes the microstructure from coarse Widmanstätten and blocky ferrite to acicular ferrite with pearlite and finer bainite, raising tensile strength and elongation. The 400°C condition demonstrates the metallurgical side; the 700°C simulation and production trial demonstrate the stress side. The combination is why the crack rate fell.
Control levers follow from this. Shell temperature at pour is the first-order variable. Metal superheat should be kept low because higher pouring temperature increases contraction strain and stress; 1560°C was preferred over 1580–1600°C, which also increased shrinkage. Pouring rate should be set for smooth filling, but it is not a major stress lever. Riser and transition design should avoid isolated hot spots and maintain feed paths. Inclusion control matters because cleavage initiation occurred at C/O/Mn-rich nonmetallic regions.
The analysis has limits. The stress values come from a finite-element model with calculated thermophysical properties and assumed interface conditions; different property sets can shift absolute peaks, so the ranking of shell temperature > metal temperature > pouring rate is more robust than any single number. The microstructure–property data apply to ZG25MnCrNiMo and this mount geometry. The production result is a reduced crack rate, not a full statistical qualification. Transfer to other section sizes, shell systems, or heat treatments requires revalidation.