I focused my study on the precision casting of a heavy-haul railway locomotive coupler made from E-grade cast steel ZG25MnCrNiMo. This component transfers traction and impact loads between vehicles, so it must combine high strength, good toughness, fatigue resistance, and internal soundness. The coupler is a hollow, thin-to-thick walled part with a complex load path, and its production by precision casting is sensitive to shrinkage porosity, hot tearing, and non-uniform grain structure. My aim was to establish a robust precision casting window that suppresses internal porosity and cracking while promoting a fine, dense solidification structure with mechanical properties above the railway standard.
I treated the coupler as a precision casting problem in which filling, solidification, stress evolution, and grain growth are strongly coupled. Instead of relying only on trial-and-error, I combined a Taguchi-style orthogonal simulation experiment, thermal and mechanical finite-element analysis, cellular-automaton grain simulation, and final trial production. The process variables I selected were pouring temperature, shell preheating temperature, and pouring time, because these variables control superheat, cooling rate, feeding ability, and thermal gradients in precision casting. I used the volume of shrinkage porosity and cavitation, together with peak stress at critical nodes, as optimization responses. The optimized precision casting parameters were then validated by temperature-field, stress-field, and microstructure simulations, followed by X-ray inspection, metallography, and tensile testing of trial couplers.
1. Coupler Geometry, Material, and Precision Casting Requirements
The coupler has an overall envelope of approximately 594 mm × 370 mm × 350 mm and a mass near 70 kg. Its internal cavity and variable section produce a wide range of local modulus values. The main functional regions include the base, the shank, the head, the knuckle, and the uncoupling cylinder. The wall thickness changes abruptly at the shoulder, at the head-to-knuckle transition, and near the base, creating natural hot spots during solidification. In precision casting, these hot spots must be fed through an intentional thermal gradient, or they become sites for shrinkage cavities, porosity, and cracking.
I used a vertical top-pouring gating system with two side branches and an open riser at the top of the head. This arrangement was chosen to promote sequential solidification from the remote thin sections toward the riser and the gating system. The side branches assist feeding at the shoulder, while the open riser compensates for liquid shrinkage in the last-solidifying region. Because the coupler is a safety-critical precision casting, the gating design must not only fill the cavity without cold shuts or misruns but also establish a temperature field that keeps the critical knuckle and shoulder regions under compression or low tension during cooling.
The material is an E-grade cast steel, ZG25MnCrNiMo, with a nominal composition designed for high strength, wear resistance, and low-temperature impact toughness. I used a charge composition with C, Si, Mn, Cr, Ni, and Mo controlled at the levels shown in Table 1. P and S were kept to a low level to reduce hot cracking sensitivity and segregation. The alloying elements increase hardenability and temper resistance, but they also alter the solidification interval and thermal conductivity, so the precision casting process must be tuned to the specific thermophysical behavior of this steel.
| Element | C | Si | Mn | Cr | Ni | Mo | P | S |
|---|---|---|---|---|---|---|---|---|
| wt.% | 0.26 | 0.45 | 1.40 | 0.55 | 0.45 | 0.25 | ≤0.35 | ≤0.35 |
I calculated the temperature-dependent thermophysical properties of the steel using a thermodynamic database with a mixed model. The properties included thermal conductivity, density, enthalpy, solid fraction, and latent heat release. These properties were imported into the precision casting simulation so that the filling and solidification predictions reflected the actual alloy rather than a generic steel. The melting range of ZG25MnCrNiMo is approximately 1,500 °C, which set the lower bound for a feasible pouring temperature. For precision casting, the superheat must be high enough to avoid premature freezing in the thin sections but low enough to limit grain coarsening and hot tearing.
2. Simulation Models and Initial Conditions
I constructed a three-dimensional finite-element model of the coupler, gating system, and shell. The mesh was refined in the casting body and coarsened in the gating and riser regions to balance accuracy and computation time. The initial simulation conditions are summarized in Table 2. These values were used as the baseline before the orthogonal optimization. The baseline pouring temperature was 1,550 °C, the shell preheating temperature was 400 °C, and the pouring time was 15 s. The interfacial heat transfer coefficient between the shell and the casting was 500 W·m⁻²·K⁻¹, while the shell-to-air coefficient was 10 W·m⁻²·K⁻¹.
| Parameter | Value |
|---|---|
| Pouring temperature, T₁ | 1,550 °C |
| Shell preheating temperature, T₂ | 400 °C |
| Ambient temperature, T₀ | 20 °C |
| Pouring time, t | 15 s |
| Shell-to-casting heat transfer coefficient, h₁ | 500 W·m⁻²·K⁻¹ |
| Shell-to-air heat transfer coefficient, h₂ | 10 W·m⁻²·K⁻¹ |
| Mesh size for gating system and riser | 8 mm |
| Mesh size for casting body | 5 mm |
The thermal field during precision casting was governed by transient heat conduction with latent heat release:
$$
\rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \rho L \frac{\partial f_s}{\partial t}
$$
where \(\rho\) is density, \(c_p\) is specific heat capacity, \(T\) is temperature, \(t\) is time, \(k\) is thermal conductivity, \(L\) is latent heat, and \(f_s\) is solid fraction. The solid fraction evolution controls not only feeding resistance but also the development of thermal stress. In my precision casting analysis, I coupled this thermal model to a mechanical model to estimate where hot cracks are most likely to initiate.
For stress evolution, I used a thermo-elasto-plastic formulation. Mechanical equilibrium was solved as:
$$
\nabla \cdot \boldsymbol{\sigma} + \mathbf{F} = \mathbf{0}
$$
where \(\boldsymbol{\sigma}\) is the stress tensor and \(\mathbf{F}\) is the body force vector. The total strain was decomposed into elastic, plastic, and thermal components:
$$
\boldsymbol{\varepsilon} = \boldsymbol{\varepsilon}^{e} + \boldsymbol{\varepsilon}^{p} + \boldsymbol{\varepsilon}^{th}
$$
The thermal strain was calculated from the temperature change relative to the initial condition:
$$
\boldsymbol{\varepsilon}^{th} = \alpha (T – T_0)\mathbf{I}
$$
where \(\alpha\) is the coefficient of thermal expansion, \(T_0\) is the reference temperature, and \(\mathbf{I}\) is the identity tensor. The elastic stress was obtained from the constitutive relation:
$$
\boldsymbol{\sigma} = \mathbf{C} : \boldsymbol{\varepsilon}^{e}
$$
For the plastic response of the E-grade steel, I used a bilinear hardening model. This model is simple enough for large-scale precision casting simulation but still captures the transition from elastic to plastic deformation:
$$
\sigma =
\begin{cases}
E_1 \varepsilon, & \varepsilon \le \varepsilon_s \\
\sigma_{0.2} + E_2(\varepsilon – \varepsilon_s), & \varepsilon > \varepsilon_s
\end{cases}
$$
Here, \(E_1\) is the elastic modulus, \(E_2\) is the hardening modulus, \(\sigma_{0.2}\) is the yield strength, and \(\varepsilon_s\) is the yield strain. I restricted the hot-cracking assessment to the mushy zone, because hot tears form when the solid fraction is high enough to form a coherent network but liquid films still exist along grain boundaries. In that fragile temperature range, even moderate tensile stress can cause irreversible separation.
For grain structure prediction, I used a cellular automaton finite-element model coupled with the Kurz-Giovanola-Trivedi growth model. The nucleation density was described by a Gaussian distribution:
$$
\frac{dn}{d(\Delta T)} =
\frac{n_{\max}}{\sqrt{2\pi}\,\Delta T_\sigma}
\exp\left[-\frac{1}{2}\left(\frac{\Delta T – \Delta T_{\max}}{\Delta T_\sigma}\right)^2\right]
$$
where \(n\) is grain density, \(\Delta T\) is undercooling, \(\Delta T_{\max}\) is the mean nucleation undercooling, \(\Delta T_\sigma\) is the standard deviation of nucleation undercooling, and \(n_{\max}\) is the maximum nucleation density. The dendritic tip growth was approximated by a KGT-type kinetic relation:
$$
v = \alpha \Delta T^2 + \beta \Delta T^3
$$
where \(v\) is the growth velocity and \(\alpha\), \(\beta\) are kinetic coefficients. In the full CAFE implementation, the local undercooling is also related to the solute diffusion field and the Gibbs-Thompson effect:
$$
\Delta T = \frac{v}{\mu} + \frac{2\Gamma}{r}
$$
where \(\mu\) is the interface kinetic coefficient, \(\Gamma\) is the Gibbs-Thompson coefficient, and \(r\) is the dendrite tip radius. The nucleation and growth parameters I used for ZG25MnCrNiMo are listed in Table 3. These parameters allowed me to predict the columnar-to-equiaxed transition and the final grain-size distribution in the precision casting.
| Element | m | k | Γ (×10⁻⁷ K·m) | D₁ (×10⁻⁹ m²·s⁻¹) |
|---|---|---|---|---|
| C | −83.024 2 | 0.166 396 | 3 | 3 |
| Cr | −1.837 04 | 0.907 287 | 3 | 3 |
| Mn | −5.172 83 | 0.737 152 | 3 | 3 |
| Mo | −2.613 29 | 0.781 625 | 3 | 3 |
| Ni | −3.872 22 | 0.803 188 | 3 | 3 |
The nucleation parameters were \(n_{s,\max}=1\times10^{9}\ \mathrm{m^{-3}}\), \(n_{v,\max}=1\times10^{11}\ \mathrm{m^{-3}}\), \(\Delta T_{s,\max}=\Delta T_{v,\max}=8\ \mathrm{K}\), and \(\Delta T_{s,\sigma}=\Delta T_{v,\sigma}=2\ \mathrm{K}\). Here, the subscripts \(s\) and \(v\) denote surface and volume nucleation, respectively. These values were selected to reproduce the observed grain morphology in the precision casting trials.
3. Orthogonal Experiment Design for Precision Casting
I designed an orthogonal experiment to minimize two objectives: the total volume of shrinkage porosity and cavitation, denoted \(V_{SP}\), and the peak stress at selected critical nodes, denoted \(\sigma\). The three factors were pouring temperature \(A\), shell preheating temperature \(B\), and pouring time \(C\). Each factor was assigned four levels. The factor levels are shown in Table 4. The range of pouring temperature was 1,530–1,590 °C, the shell preheating range was 350–500 °C, and the pouring time range was 28–34 s. These ranges were chosen around the practical precision casting window for this alloy and part geometry.
| Level | A: Pouring temperature (°C) | B: Shell preheating temperature (°C) | C: Pouring time (s) |
|---|---|---|---|
| 1 | 1,530 | 350 | 28 |
| 2 | 1,550 | 400 | 30 |
| 3 | 1,570 | 450 | 32 |
| 4 | 1,590 | 500 | 34 |
I used an \(L_{16}(4^3)\) Taguchi array, which allowed me to evaluate the main effects of the three factors with a manageable number of precision casting simulations. For each run, I recorded the shrinkage porosity volume, the peak stress at the critical nodes, and the corresponding signal-to-noise ratios. For a smaller-is-better response, the signal-to-noise ratio was calculated as:
$$
\eta = -10 \log_{10}\left(\frac{1}{n}\sum_{i=1}^{n} y_i^2\right)
$$
where \(y_i\) is the response value and \(n\) is the number of observations. The average signal-to-noise ratio for each factor level was:
$$
\bar{\eta}_j = \frac{1}{m}\sum_{i=1}^{m} \eta_{j,i}
$$
The range of each factor was then obtained from:
$$
R_j = \max_i(\bar{\eta}_{j,i}) – \min_i(\bar{\eta}_{j,i})
$$
A larger range indicates a stronger effect on the response. The orthogonal array and results are summarized in Table 5. The shrinkage porosity volume ranged from 0.887 cm³ to 1.252 cm³, while the critical-node stress ranged from 343.366 MPa to 353.024 MPa. The signal-to-noise ratios for both responses were used to identify the most influential factors and the best levels.
| Run | A (°C) | B (°C) | C (s) | VSP (cm³) | S/N for VSP (dB) | σ (MPa) | S/N for σ (dB) |
|---|---|---|---|---|---|---|---|
| 1 | 1,530 | 350 | 28 | 1.048 | −0.407 | 346.124 | −51.641 |
| 2 | 1,530 | 400 | 30 | 0.887 | 0.257 | 344.063 | −51.762 |
| 3 | 1,530 | 450 | 32 | 0.910 | 0.436 | 345.167 | −51.881 |
| 4 | 1,530 | 500 | 34 | 1.082 | 0.128 | 347.598 | −51.999 |
| 5 | 1,550 | 350 | 30 | 0.894 | 0.284 | 344.913 | −51.952 |
| 6 | 1,550 | 400 | 32 | 1.021 | −0.587 | 347.017 | −51.959 |
| 7 | 1,550 | 450 | 34 | 1.054 | 0.104 | 346.577 | −51.998 |
| 8 | 1,550 | 500 | 28 | 1.089 | −0.011 | 343.366 | −52.056 |
| 9 | 1,570 | 350 | 32 | 1.108 | −0.118 | 353.024 | −52.032 |
| 10 | 1,570 | 400 | 34 | 0.990 | −0.098 | 349.998 | −52.040 |
| 11 | 1,570 | 450 | 28 | 1.124 | −0.190 | 352.604 | −52.064 |
| 12 | 1,570 | 500 | 30 | 1.252 | −0.367 | 352.572 | −52.104 |
| 13 | 1,590 | 350 | 34 | 0.897 | −0.279 | 345.059 | −52.095 |
| 14 | 1,590 | 400 | 28 | 1.125 | −0.337 | 352.233 | −52.101 |
| 15 | 1,590 | 450 | 30 | 1.074 | −0.356 | 347.722 | −52.122 |
| 16 | 1,590 | 500 | 32 | 0.900 | −0.287 | 345.979 | −52.156 |
The variance analysis showed that pouring temperature had a statistically significant effect on both shrinkage porosity and critical-node stress. Shell preheating temperature and pouring time had much weaker effects on stress within the chosen range. For shrinkage porosity, the influence ranking was \(A > C > B\). For critical-node stress, the ranking was also \(A > C > B\). This indicates that the thermal state established by the pouring temperature dominates the precision casting outcome, while the pouring time plays a secondary role by changing the filling temperature distribution. The shell preheating temperature is important for fluidity and thermal gradient control, but its effect on the final porosity and stress is smaller than that of pouring temperature in this specific coupler geometry.
Using the signal-to-noise ratios, I identified the best level combination for each objective. For minimum shrinkage porosity, the best combination was \(A_3B_4C_1\): pouring temperature 1,570 °C, shell preheating temperature 500 °C, and pouring time 28 s. For minimum critical-node stress, the best combination was \(A_3B_1C_2\): pouring temperature 1,570 °C, shell preheating temperature 350 °C, and pouring time 30 s. Because the two objectives favored different shell preheating temperatures and slightly different pouring times, I took the compromise between the two optima. The final precision casting parameters were pouring temperature 1,570 °C, shell preheating temperature 425 °C, and pouring time 29 s. These parameters are summarized in Table 6.
| Objective | Best combination | Pouring temperature (°C) | Shell preheating temperature (°C) | Pouring time (s) |
|---|---|---|---|---|
| Minimum VSP | A₃B₄C₁ | 1,570 | 500 | 28 |
| Minimum σ | A₃B₁C₂ | 1,570 | 350 | 30 |
| Compromise for precision casting | A₃B(avg)C(avg) | 1,570 | 425 | 29 |
4. Simulation Validation of the Optimized Precision Casting Parameters
4.1 Temperature Field and Solidification Sequence
I re-ran the precision casting simulation with the optimized parameters to examine the temperature field, solidification sequence, shrinkage porosity, stress, and grain structure. The temperature-field results showed that solidification began in the middle and lower base regions, then progressed outward and upward, and finally ended at the shoulder and gate regions. This sequence is consistent with the local wall thickness and the gating system design. The thin sections lost heat quickly and solidified first, while the thicker shoulder and gating regions remained liquid longer. This behavior is desirable in precision casting because it encourages feeding from the gating system toward the last-solidifying hot spots.
The coupler has large section changes at the head-to-knuckle transition and at the shoulder. If these regions solidify before the gating system can feed them, shrinkage cavities form. I placed side gates at the shoulder to provide local feeding and to reduce the temperature gradient across the section change. The open riser at the top of the head provided additional liquid metal for the last stage of solidification. As a result, the final shrinkage porosity was confined to the top of the head and the belly of the uncoupling cylinder. The total predicted shrinkage volume under the optimized precision casting parameters was 0.879 cm³, which is small compared with the total casting volume. More importantly, no shrinkage porosity was predicted in the critical knuckle or shoulder load-bearing sections.
4.2 Stress Field and Hot Cracking Tendency
I analyzed the stress field at five representative nodes in the coupler. The nodes were located in the knuckle interior, the head, the shoulder, the bottom of the base, and the shoulder gate region. The solidification and stress histories of these nodes revealed how the local cooling rate and feeding conditions affect hot tearing. The knuckle interior solidified early and reached a coherent solid network before the surrounding thick sections. The base node solidified relatively slowly, and because it was constrained by the adjacent thick sections and the gating system, it developed the highest tensile stress. The shoulder gate node solidified slowly because of the thick runner, but it was also able to contract more freely, so its stress remained lower.
The peak stress in the bottom region reached approximately 470 MPa at about 6,000 s. The maximum stress at the critical coupler locations was approximately 445 MPa. These values are below the tensile strength of the as-cast steel, but hot tearing is not governed only by the final tensile strength. In the mushy zone, the material has very low ductility, and a hot crack can initiate when the local tensile strain or stress exceeds a critical value. I therefore used a hot-cracking index:
$$
HCI = \frac{\sigma_{\max}}{\sigma_c}
$$
where \(\sigma_{\max}\) is the maximum principal stress in the vulnerable temperature range and \(\sigma_c\) is the critical cohesive strength of the mushy zone. The predicted hot-cracking index remained below the critical threshold in the knuckle and shoulder. The areas with the highest hot-cracking tendency were the shoulder-to-uncoupling-cylinder transition and the base-to-gating junction. These are thick regions with local heat sources, so their solidification is delayed and their contraction is constrained. However, the optimized precision casting parameters reduced the thermal gradient enough to keep the hot-cracking risk low.
| Node | Location | Solidification behavior | Peak stress or trend | Interpretation |
|---|---|---|---|---|
| 69754 | Knuckle interior | Solidifies early; reaches coherent network before surrounding thick sections | Moderate | Fast cooling, fine structure, low feeding demand |
| 29772 | Head region | Intermediate solidification | Moderate | Riser provides local feeding |
| 51539 | Shoulder region | Intermediate solidification | Moderate | Side gate reduces thermal gradient |
| 59684 | Bottom of base | Slow solidification; constrained by thick sections | Approximately 470 MPa at 6,000 s | Highest stress; but below critical hot-cracking threshold |
| 35369 | Shoulder gate | Latest solidification; influenced by thick runner | Approximately 200 MPa | Slow cooling but relatively free contraction |
The stress simulation confirmed that the optimized precision casting parameters did not eliminate all thermal gradients, but they shifted the gradients away from the most critical load-bearing sections. The highest stress remained in the base region, which is less sensitive to hot tearing because it is thicker and less constrained in the final stage of cooling. The shoulder and knuckle, which carry the main service loads, experienced lower stress and a lower hot-cracking index. This is a key benefit of the orthogonal optimization: it does not simply reduce the average stress but redistributes the thermal history so that the critical precision casting sections remain sound.
4.3 Microstructure Simulation
I used the CAFE model to predict the grain structure under the optimized precision casting parameters. The simulation showed three characteristic zones: a fine-grained surface zone, a columnar zone adjacent to the shell wall, and an equiaxed zone in the interior. The surface zone had the highest undercooling because the liquid metal contacted the relatively cold shell, so nucleation was intense and grain growth was limited. The columnar zone formed where the thermal gradient was steep and the growth direction was roughly perpendicular to the shell wall. The equiaxed zone formed in the central and thicker regions where the thermal gradient decreased and constitutional undercooling allowed new grains to nucleate ahead of the columnar front.
The grain structure was not uniform throughout the coupler. In the head cross-section, the wall was relatively thin, so the columnar zone occupied a larger fraction of the section and the central equiaxed zone was smaller. In the shoulder, the edges showed more columnar grains, while the central region had more equiaxed grains. At the intersection of the head and knuckle and at the shoulder gate, the solidification time was longer, so the equiaxed grains were coarser than in the surrounding regions. However, the knuckle and shoulder, which are the main load-bearing regions, had a high fraction of fine equiaxed grains. The predicted grain-size grade was in the range of 1–3, indicating a relatively fine precision casting structure.
The grain density and grain size were estimated from the nucleation and growth results using:
$$
N_v = \int_{0}^{\Delta T} \frac{dn}{d(\Delta T)} d(\Delta T)
$$
and
$$
d = \left(\frac{6}{\pi N_v}\right)^{1/3}
$$
where \(N_v\) is the volumetric grain density and \(d\) is the equivalent grain diameter. The simulation showed that the optimized pouring temperature of 1,570 °C provided enough superheat to fill the thin sections without causing excessive grain coarsening. The shell preheating temperature of 425 °C reduced the thermal shock at the mold wall, which lowered the columnar-to-equiaxed transition temperature and promoted equiaxed grain formation. The pouring time of 29 s was long enough to avoid turbulent filling and cold shuts, but short enough to maintain a favorable temperature gradient toward the riser.
5. Trial Precision Casting, X-ray Inspection, and Microstructure
I produced trial couplers using the optimized precision casting parameters. The external surfaces were smooth and complete, with no visible surface cracks, cold shuts, or macro-porosity. I selected one trial casting and sectioned it at the head, the knuckle joint, and the base. X-ray inspection was performed on these sections. The results are summarized in Table 8. No internal cracks or porosity were detected in any of the inspected regions. This confirmed that the optimized precision casting process produced a sound internal structure in the critical load-bearing sections.
| Inspected region | Result | Interpretation |
|---|---|---|
| Head | No cracks or porosity detected | Riser and gating design provided adequate feeding |
| Knuckle joint | No cracks or porosity detected | Critical load path is sound |
| Base | No cracks or porosity detected | Thermal gradient was controlled despite thick section |
I then examined the microstructure of the as-cast coupler and the heat-treated coupler. The as-cast microstructure consisted of primary ferrite and pearlite. The ferrite appeared as coarse lamellae or blocky regions, with a maximum length on the order of 100 μm and a maximum thickness on the order of 10 μm. The pearlite was closely connected to the ferrite, and the distribution was generally uniform. Energy-dispersive spectroscopy showed that C and Mn were uniformly distributed around the ferrite, with no large-scale macro-segregation. This indicates that the solidification conditions in the precision casting were stable and that the alloying elements were well distributed.
The heat treatment consisted of normalizing, quenching at 910 °C for 2 h, and tempering at 590 °C for 3 h. After heat treatment, the microstructure transformed to tempered sorbite. The tempered sorbite was fine and uniform, with grain sizes below 1 μm. The presence of Cr, Mo, and Mn increased the hardenability and retarded the growth of austenite during quenching. Si, Mn, and Ni improved the temper resistance, so the martensitic structure did not coarsen significantly during high-temperature tempering. The final structure retained a fine morphology, which contributed to the high strength and good ductility of the precision casting.
| Condition | Main phases | Scale | Morphology |
|---|---|---|---|
| As-cast | Ferrite + pearlite | Ferrite length up to about 100 μm; thickness up to about 10 μm | Coarse lamellar/blocky ferrite connected with pearlite |
| Heat-treated | Tempered sorbite | Grain size below 1 μm | Fine, uniform, and well distributed |
6. Mechanical Properties of the Precision Cast Coupler
I evaluated the mechanical properties of the coupler in both the as-cast and heat-treated conditions. Specimens were taken from the shoulder regions, with four specimens from one area and two from another, giving a total of six specimens per condition. The tensile tests showed that the as-cast material had an average tensile strength of 625 MPa and an elongation of about 2.3–2.7%. After heat treatment, the average tensile strength reached 1,020 MPa, and the elongation increased to 14.5%. The yield strength was 810 MPa, the reduction of area was 34.5%, and the elastic modulus was 215,470 MPa. The hardness after heat treatment was 315 HBW. These values exceed the railway standard requirements, as shown in Table 10.
| Condition | σ0.2 (MPa) | σb (MPa) | δ (%) | ψ (%) | E (MPa) | Hardness (HBW) |
|---|---|---|---|---|---|---|
| Standard requirement | ≥590 | ≥830 | ≥14 | ≥30 | 174,000 | 241–311 |
| As-cast | — | 625 | 2.3–2.7 | — | — | 238 |
| Heat-treated | 810 | 1,020 | 14.5 | 34.5 | 215,470 | 315 |
The improvement in mechanical properties is directly related to the microstructure. In the as-cast condition, the coarse ferrite and pearlite structure provides limited resistance to plastic deformation. After quenching and tempering, the tempered sorbite provides a fine dispersion of carbides in a ferrite matrix, which increases both strength and toughness. The high hardenability of ZG25MnCrNiMo ensures that the thick sections also transform to martensite during quenching, so the heat treatment is effective throughout the precision casting. The fine grain size produced by the optimized precision casting process further contributes to the strength and ductility by increasing the number of grain boundaries that impede dislocation motion and crack propagation.
The tensile results also confirm that the optimized precision casting process did not introduce hidden defects that could reduce load-bearing capacity. If shrinkage porosity or hot cracks had been present in the knuckle or shoulder, the tensile specimens would have shown premature failure and low elongation. Instead, the heat-treated material achieved an average tensile strength of 1,020 MPa and an elongation of 14.5%, which indicates a sound and well-heat-treated precision casting.
7. Discussion
The orthogonal precision casting study showed that pouring temperature is the dominant variable for both shrinkage porosity and stress in this coupler. A pouring temperature of 1,570 °C provided a balance between fluidity and solidification rate. Lower pouring temperatures increased the risk of cold shuts and incomplete filling in the thin sections, while higher pouring temperatures increased the liquid shrinkage and the thermal gradient, which promoted porosity and hot tearing. The optimized pouring temperature was therefore not simply the highest or lowest value, but the level that produced the best compromise between filling and feeding.
The shell preheating temperature also played an important role, even though its statistical effect was weaker than that of pouring temperature. A shell preheating temperature of 425 °C reduced the heat loss from the liquid metal during filling and slowed the cooling rate near the mold wall. This improved feeding and reduced the columnar grain fraction. If the shell is too cold, the liquid metal freezes prematurely and the thin sections may not fill. If the shell is too hot, the solidification time becomes too long, the grain size increases, and the hot spots remain liquid for an extended period, which can increase shrinkage porosity. The compromise value of 425 °C was therefore selected to promote a fine equiaxed structure without sacrificing feeding.
The pouring time of 29 s was also a compromise. A shorter pouring time reduces heat loss during filling and helps maintain a high temperature in the last-filled regions, but it can also increase turbulence and mold erosion. A longer pouring time reduces turbulence but allows more heat loss, which can cause premature freezing in the thin sections. The optimized pouring time provided smooth filling and a favorable temperature gradient from the thin sections toward the riser. This is especially important in precision casting, where the shell is often preheated and the metal must fill a complex, thin-walled cavity without defects.
The stress and hot-cracking analysis showed that the optimized precision casting parameters did not eliminate thermal stress entirely. The base region still experienced the highest stress because of its thick section and constrained contraction. However, the critical knuckle and shoulder regions remained below the hot-cracking threshold. This outcome is consistent with the design intent: the gating system and riser were placed to feed the thick sections and to shift the last-solidifying regions away from the most highly stressed load paths. The use of side gates at the shoulder was particularly effective because it reduced the thermal gradient across the section change and provided local liquid metal for feeding.
The microstructure simulation and experimental observations were in good agreement. Both showed a fine surface zone, a columnar zone near the shell, and an equiaxed zone in the interior. The predicted grain-size grade of 1–3 matched the fine structure observed after heat treatment. The heat-treated tempered sorbite had grain sizes below 1 μm, which is much finer than the as-cast ferrite and pearlite. This fine structure is the main reason for the high tensile strength and good elongation. The uniform distribution of C and Mn also indicates that the precision casting process did not produce significant macro-segregation, which is important for consistent mechanical properties.
From a manufacturing perspective, the optimized precision casting parameters are practical and reproducible. The pouring temperature of 1,570 °C is within the normal range for E-grade cast steel. The shell preheating temperature of 425 °C is achievable with standard investment casting furnaces. The pouring time of 29 s is compatible with the gating system and the production cycle. The trial castings showed no internal cracks or porosity after X-ray inspection, and the heat-treated mechanical properties exceeded the standard requirements. This suggests that the orthogonal optimization approach can be applied to other precision casting components with similar geometry and alloy systems.
Nevertheless, the study also shows that precision casting of heavy-section couplers remains sensitive to local geometry. The base region, the shoulder-to-cylinder transition, and the head top are natural hot spots. Even with optimized parameters, these regions require careful gating and riser design. Future work could include adaptive control of shell preheating, real-time monitoring of pouring temperature, and localized chilling or insulation to further reduce the thermal gradient. The combination of simulation, orthogonal design, and trial production used here provides a reliable framework for developing precision casting processes for safety-critical railway components.
8. Conclusions
I optimized the precision casting process for a steel locomotive coupler using an orthogonal simulation experiment with three factors and four levels. The optimized parameters were a pouring temperature of 1,570 °C, a shell preheating temperature of 425 °C, and a pouring time of 29 s. These parameters were obtained by balancing the minimum shrinkage porosity volume and the minimum critical-node stress.
Under the optimized precision casting parameters, the solidification sequence started in the middle and lower base, progressed outward, and ended at the shoulder and gate. The final shrinkage porosity was confined to the head top and the uncoupling cylinder belly, with a total predicted volume of 0.879 cm³. No shrinkage porosity was predicted in the critical knuckle or shoulder sections. The maximum stress at the critical locations was approximately 445 MPa, and the hot-cracking tendency remained low.
The microstructure simulation predicted a fine surface zone, a columnar zone, and an interior equiaxed zone. The grain-size grade was in the range of 1–3. Trial precision castings showed no internal cracks or porosity under X-ray inspection. The as-cast microstructure consisted of ferrite and pearlite, while the heat-treated microstructure consisted of fine tempered sorbite with grain sizes below 1 μm.
The as-cast tensile strength was 625 MPa. After normalizing, quenching at 910 °C for 2 h, and tempering at 590 °C for 3 h, the average tensile strength reached 1,020 MPa, the yield strength was 810 MPa, the elongation was 14.5%, and the reduction of area was 34.5%. These properties exceed the railway standard requirements. The results demonstrate that the optimized precision casting process produces a dense, crack-free coupler with fine grains and high mechanical performance.
Overall, the study confirms that precision casting of steel locomotive couplers can be made robust by combining orthogonal process optimization, thermal and stress simulation, and cellular-automaton microstructure prediction. The optimized parameters provide a reliable basis for producing safety-critical precision castings with consistent internal quality and mechanical properties.