I investigated the precision casting process and solidification microstructure of a steel locomotive coupler because this component must connect railway vehicles and transmit traction forces under severe service conditions. My central aim was to obtain a dense microstructure, avoid internal pores and cracks, and achieve mechanical properties above the railway standard. I used an orthogonal simulation experiment, casting simulation, stress analysis, and cellular automaton finite element microstructure modeling. Throughout this work, I also kept evaporative pattern casting as a reference process because evaporative pattern casting is another important precision casting route for complex steel parts. Although evaporative pattern casting can reduce tooling complexity in some cases, my coupler required the dimensional accuracy, surface finish, and controlled solidification behavior that investment casting can provide. Therefore, I focused on investment casting while repeatedly comparing its thermal and stress behavior with what I would expect from evaporative pattern casting.
The coupler had a hollow structure with strongly varying wall thickness. Its overall envelope was approximately 594 mm × 370 mm × 350 mm, and its mass was about 70 kg. The component included a base, a shank, a head, a knuckle, and an uncoupling cylinder. Because the coupler is a safety-critical part, internal cavities, shrinkage porosity, hot cracks, and coarse microstructures must be avoided. I designed a top-pouring gating system with side branches and an open riser at the top of the head. I used investment casting with a ceramic shell, and I optimized the pouring temperature, shell preheating temperature, and pouring time. I also evaluated how the optimized parameters influenced the temperature field, stress field, shrinkage porosity, and grain structure. In parallel, I noted that evaporative pattern casting would require different pattern removal and shelling strategies, but the same fundamental solidification principles would still govern defect formation.
The alloy was an E-grade cast steel, ZG25MnCrNiMo. Its composition gave good wear resistance, low-temperature impact toughness, and high strength. The measured composition before pouring is shown in Table 1. I used the thermodynamic database in ProCAST with a mixed model to calculate temperature-dependent thermophysical properties such as thermal conductivity, density, and enthalpy. These properties are essential for reliable simulation of investment casting. I also compared these property trends with those I would expect in evaporative pattern casting, where the foam pattern decomposition can alter local heat transfer and gas pressure. In the present investment casting process, the ceramic shell provided a more stable thermal boundary condition than the decomposing foam pattern in evaporative pattern casting.
| Element | C | Si | Mn | Cr | Ni | Mo | P | S |
|---|---|---|---|---|---|---|---|---|
| Mass fraction / % | 0.26 | 0.45 | 1.40 | 0.55 | 0.45 | 0.25 | ≤0.35 | ≤0.35 |
I built a three-dimensional model of the coupler and its gating system. The mesh size was 8 mm for the gating system and riser and 5 mm for the casting body. The initial conditions for the filling and solidification simulation are listed in Table 2. The initial pouring temperature in the baseline simulation was 1 550 °C, the shell preheating temperature was 400 °C, the ambient temperature was 20 °C, and the pouring time was 15 s. The heat transfer coefficient between the shell and the casting was 500 W·m-2·K-1, and the heat transfer coefficient between the shell and air was 10 W·m-2·K-1. These values provided a stable numerical framework for the orthogonal experiment.
| Parameter | Value |
|---|---|
| Pouring temperature T1 / °C | 1 550 |
| Shell preheating temperature T2 / °C | 400 |
| Ambient temperature T0 / °C | 20 |
| Pouring time t / s | 15 |
| Shell–casting heat transfer coefficient h1 / (W·m-2·K-1) | 500 |
| Shell–air heat transfer coefficient h2 / (W·m-2·K-1) | 10 |
| Mesh size for gating system and riser / mm | 8 |
| Mesh size for casting body / mm | 5 |
For stress analysis, I used a thermo-elastic-plastic model. Because hot tearing mainly occurs in the mushy zone, I concentrated the stress analysis on the region between the solidus and liquidus. I described the material behavior with a classical bilinear hardening model. The stress–strain relationship was written as
$$
\sigma = E_1 \varepsilon \quad (\varepsilon \leq \varepsilon_s)
$$
$$
\sigma = \sigma_{0.2} + E_2 (\varepsilon – \varepsilon_s) \quad (\varepsilon > \varepsilon_s)
$$
where σ is stress, σ0.2 is yield strength, ε is strain, εs is the strain at yield, and E1 and E2 are elastic moduli for the elastic and plastic branches. I used this model to evaluate the stress concentration at the base, shank, head, knuckle, and uncoupling cylinder. The same stress model could be applied to evaporative pattern casting if the thermal boundary conditions were adjusted for foam decomposition.
For microstructure simulation, I combined a cellular automaton with a finite element model, known as CAFE. The model included a continuous nucleation Gaussian distribution and a dendrite tip growth kinetics model, namely the KGT model. The nucleation density was expressed as
$$
\frac{dn}{d(\Delta T)} = \frac{n_{\max}}{\sqrt{2\pi}\Delta T_{\sigma}} \exp\left[-\frac{1}{2}\left(\frac{\Delta T – \Delta T_{\max}}{\Delta T_{\sigma}}\right)^2\right]
$$
where n is grain density, ΔT is undercooling, ΔTmax is the mean nucleation undercooling, ΔTσ is the standard deviation of nucleation undercooling, and nmax is the maximum nucleation density obtained by integrating the normal distribution from zero. The KGT model provided the growth kinetics of the dendrite tip. I used the kinetic parameters and nucleation Gaussian distribution parameters for ZG25MnCrNiMo listed in Table 3. These parameters allowed me to predict equiaxed and columnar grain structures in the coupler. I also considered how the same CAFE approach would need to be modified for evaporative pattern casting, because the foam pattern changes the local cooling rate and gas evolution.
| Parameter | C | Cr | Mn | Mo | Ni |
|---|---|---|---|---|---|
| Liquidus slope m | −83.024 2 | −1.837 04 | −5.172 83 | −2.613 29 | −3.872 22 |
| Partition coefficient k | 0.166 396 | 0.907 287 | 0.737 152 | 0.781 625 | 0.803 188 |
| Gibbs–Thompson coefficient Γ / (×10−7 K·m) | 3 | 3 | 3 | 3 | 3 |
| Liquid diffusion coefficient D1 / (×10−9 m2·s−1) | 3 | 3 | 3 | 3 | 3 |
| Nucleation parameters |
ns,max = 1×109 m−3, nv,max = 1×1011 m−3, ΔTs,max = ΔTv,max = 8 K, ΔTs,σ = ΔTv,σ = 2 K |
||||
I designed a Taguchi L16(43) orthogonal experiment to minimize the shrinkage porosity volume, VSP, and the stress, σ, at critical nodes. The three factors were pouring temperature (A), shell preheating temperature (B), and pouring time (C). The baseline experience values were 1 580 °C, 400 °C, and 30 s, respectively. Because the liquidus of ZG25MnCrNiMo was about 1 500 °C, I selected a pouring temperature range of 1 530–1 590 °C. I selected a shell preheating temperature range of 350–500 °C and a pouring time range of 28–34 s. The factor levels are shown in Table 4, and the orthogonal experiment design and results are shown in Table 5. I calculated the signal-to-noise ratio, S/N, for both VSP and σ. The smaller-the-better criterion was used because lower porosity and lower stress are preferred. I also compared this optimization strategy with what I would use for evaporative pattern casting; in evaporative pattern casting, the pouring time and pattern decomposition rate often interact strongly, so a similar orthogonal approach would be valuable.
| 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 |
| 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 significant effect on both shrinkage porosity volume and critical stress. Shell preheating temperature and pouring time had only a small effect on the critical stress. The range analysis for VSP showed that the influence order was A > C > B. The optimal combination for minimum VSP was A3B4C1, that is, pouring temperature 1 570 °C, shell preheating temperature 500 °C, and pouring time 28 s. The range analysis for σ showed that the influence order was also A > C > B. The optimal combination for minimum σ was A3B1C2, that is, pouring temperature 1 570 °C, shell preheating temperature 350 °C, and pouring time 30 s. Because I had two objectives, I averaged the two optimal conditions. The final optimized investment casting parameters were a pouring temperature of 1 570 °C, a shell preheating temperature of 425 °C, and a pouring time of 29 s. I used these parameters for all subsequent temperature, shrinkage, stress, and microstructure simulations. I also noted that if I had used evaporative pattern casting, the optimal pouring time might shift because the foam pattern would decompose and generate gas, but the optimal pouring temperature would likely remain in a similar range.
| Response | Factor A | Factor B | Factor C | Optimal combination |
|---|---|---|---|---|
| Shrinkage porosity volume VSP | Significant | Small | Moderate | A3B4C1: 1 570 °C, 500 °C, 28 s |
| Critical stress σ | Significant | Small | Moderate | A3B1C2: 1 570 °C, 350 °C, 30 s |
| Averaged optimized condition | 1 570 °C | 425 °C | 29 s | A3B2.5C1.5 |
After determining the optimized parameters, I analyzed the temperature field during solidification. The cooling rate gradually decreased as solidification progressed. The casting first solidified from the central and lower base regions, then solidified toward the surrounding areas, and finally the shoulder and gate regions solidified. This solidification sequence matched the wall thickness distribution and the gating system design. Because the coupler had a large wall thickness variation, I used the gating system to control the solidification sequence and reduce the temperature gradient. The head, knuckle, and shoulder connection had abrupt cross-section changes. I placed gates on both sides of the shoulder to improve feeding and avoid shrinkage cavities and porosity. In evaporative pattern casting, the same abrupt sections would also require careful feeding, but the foam pattern could impose additional back pressure and alter the filling sequence.
The shrinkage porosity simulation under the optimized parameters showed that the total shrinkage porosity volume was about 0.879 cm3. The pores were only located at the top of the head and in the belly of the uncoupling cylinder. The critical working regions of the coupler had no shrinkage porosity. This result confirmed that the optimized parameters provided adequate feeding. I also compared this result with typical evaporative pattern casting defects; in evaporative pattern casting, shrinkage porosity often appears near the last solidified regions because the foam pattern can absorb heat and delay solidification. In my investment casting process, the ceramic shell and the top riser allowed more predictable feeding.
I then analyzed the stress field. I selected five nodes at locations that would be prone to fatigue during service. The results showed that the node inside the knuckle started to solidify first. At 3 500 s, its solid fraction became similar to that of the other nodes, and the solidification rate decreased. The node at the bottom of the coupler solidified relatively slowly. Because solidification shrinkage was not synchronized, the maximum stress reached about 470 MPa at 6 000 s. The node at the shoulder gate finished solidification at about 5 000 s because it was affected by the thick gate. This node solidified most slowly, but because the geometry was simple and contraction was relatively free, its contraction stress was lower, close to 200 MPa. The hot cracking index indicated that the abrupt cross-section regions had a higher hot tearing tendency. The shoulder-to-uncoupling-cylinder junction and the base-to-gating-system junction were relatively thick and contained a heat source from the gating system. These regions solidified later, and their contraction was constrained, which increased the thermal stress. However, the stress at the critical nodes remained far below the tensile strength of the material. Therefore, the risk of hot cracks was low. I also considered that evaporative pattern casting could produce different stress states because the foam pattern decomposition changes the local cooling rate, but the same principle of avoiding hot spots would apply.
The microstructure simulation showed three distinct zones: an equiaxed grain zone in the interior, a columnar grain zone near the wall and perpendicular to the mold surface, and a fine grain zone at the casting surface. The equiaxed and columnar grains in thick sections were relatively coarse. The casting surface had a larger undercooling, so its nucleation ability was stronger, and its grains were finer and denser. For E-grade steel, it is important to reduce columnar grains and promote equiaxed grains. When the composition is fixed, grain morphology control depends on the temperature field, which directly affects the equiaxed grain fraction. In the head cross-section, the wall was thinner, so the columnar grain fraction was higher and the central equiaxed zone was smaller. In the shoulder region, the upper and lower edges had more columnar grains and fewer equiaxed grains. At the intersection of the head and knuckle and at the shoulder gate, the equiaxed grains were more developed because the solidification time was longer and the grains were coarser than in the surrounding areas. The other regions cooled faster, so their grains were finer and more uniform. Overall, the knuckle and shoulder regions, which carry longitudinal forces, had fine and uniform microstructures. The fine grain zone and equiaxed grain zone occupied a large fraction, which provided a good foundation for further heat treatment refinement. In evaporative pattern casting, the foam pattern decomposition can also influence the local cooling rate and grain structure, so similar CAFE modeling would be useful for evaporative pattern casting optimization.
I fabricated trial couplers using the optimized investment casting parameters. The castings had smooth surfaces without macroscopic cracks or holes. I randomly selected one casting from the same batch, sectioned it, and performed X-ray inspection on the head, the knuckle connection, and the base. The X-ray results showed no cracks or shrinkage cavities. I also sampled the shoulder region of the as-cast coupler and examined it with optical microscopy, scanning electron microscopy, and energy-dispersive X-ray spectroscopy. The as-cast room-temperature microstructure consisted of primary ferrite and pearlite. The coarse lamellar primary ferrite was closely connected with the pearlite formed by the eutectoid reaction. The maximum ferrite grain length was about 100 μm, and the maximum thickness was about 10 μm. Under high-magnification SEM, the bright ferrite regions and gray pearlite regions had clear boundaries. The ferrite and pearlite sizes were similar and uniformly distributed. The main alloying elements were consistent with the pre-pour composition. Carbon and manganese were uniformly distributed around the ferrite, and no large macroscopic segregation was observed. The microstructure was relatively uniform. After heat treatment, which consisted of normalizing, austenitizing at 910 °C for 2 h, quenching, and tempering at 590 °C for 3 h, the microstructure transformed into tempered sorbite. The tempered sorbite was fine and uniform. Because Cr, Mo, and Mn are medium-carbide and weak-carbide elements, they slow austenite grain growth during austenitizing. The alloying elements shift the continuous cooling transformation curve to the right and improve hardenability. In addition, Si, Mn, and Ni greatly improve tempering stability. During high-temperature tempering, the transformation from martensite to ferrite did not cause obvious coarsening, so the fine martensite morphology was retained. Under high-magnification SEM, the tempered sorbite grain size was less than 1 μm.
I tested the mechanical properties of the coupler before and after heat treatment. I took samples from the shoulder regions and performed tensile tests. The results are shown in Table 6. The as-cast average tensile strength was 625 MPa, and the elongation was 2.7%. After quenching and tempering, the average tensile strength reached 1 020 MPa, and the elongation reached 14.5%. The strength and ductility were both greatly improved compared with the as-cast condition. The heat-treated properties exceeded the standard requirements. I also compared these properties with what might be achieved in evaporative pattern casting. In evaporative pattern casting, the cooling rate can be lower in thick sections, which may lead to coarser grains and lower strength unless the alloy and heat treatment are carefully controlled. My investment casting process, combined with the optimized parameters and heat treatment, produced a fine tempered sorbite structure and excellent mechanical properties.
| Condition | Yield strength σ0.2 / MPa | Tensile strength σb / MPa | Elongation δ / % | Reduction of area ψ / % | Elastic modulus E / MPa | Hardness (HBW) |
|---|---|---|---|---|---|---|
| Standard | ≥590 | ≥830 | ≥14 | ≥30 | 174 000 | 241–311 |
| As-cast | — | 625 | 2.3 | — | — | 238 |
| Heat-treated | 810 | 1 020 | 14.5 | 34.5 | 215 470 | 315 |
The optimized process also reduced the risk of internal defects. The X-ray inspection confirmed that no internal cracks or pores were present. The metallographic observations were consistent with the microstructure simulation. The grain size was fine, and the mechanical properties were higher than the standard requirements. These results demonstrated that the orthogonal optimization of pouring temperature, shell preheating temperature, and pouring time produced a reasonable investment casting process for the locomotive coupler. The process also provided a useful comparison with evaporative pattern casting, because both processes aim to produce complex steel components with high integrity. In evaporative pattern casting, the pattern is made of foam, and the molten metal replaces the foam during pouring. The foam decomposes and produces gas, which can cause porosity if the gas is not removed properly. In investment casting, the pattern is removed before pouring, so the mold cavity is clean and the gas evolution is much lower. This difference is one reason I selected investment casting for the coupler. Nevertheless, I found that the thermal and stress analysis methods developed here could also be adapted to evaporative pattern casting.
I also examined the effect of alloying elements on the heat treatment response. Carbon provides strength but must be controlled to avoid excessive carbides. Manganese improves hardenability and tempering stability. Chromium and molybdenum form carbides and improve hardenability and wear resistance. Nickel improves toughness and low-temperature impact performance. The combination of these elements allowed the coupler to achieve a good balance of strength and ductility after quenching and tempering. The fine tempered sorbite structure was responsible for the high tensile strength of 1 020 MPa and the elongation of 14.5%. The hardness increased from 238 HBW in the as-cast condition to 315 HBW after heat treatment. The elastic modulus also increased to 215 470 MPa. These values are consistent with the expected behavior of a low-alloy cast steel after appropriate heat treatment. In evaporative pattern casting, similar alloy design and heat treatment would be needed to achieve comparable properties, but the casting process would require careful control of foam density, coating permeability, and pouring temperature to avoid gas defects.
From a process design perspective, the most important finding was that pouring temperature had the strongest influence on both shrinkage porosity and stress. A pouring temperature of 1 570 °C provided sufficient superheat for filling and feeding without causing excessive grain coarsening or thermal stress. The shell preheating temperature of 425 °C was a compromise between the two objectives. A higher shell preheating temperature, such as 500 °C, reduced shrinkage porosity, but it increased stress at some nodes. A lower shell preheating temperature, such as 350 °C, reduced stress but slightly increased shrinkage porosity. The pouring time of 29 s provided a reasonable filling rate. If the pouring time was too short, the filling could be turbulent and entrain gas. If the pouring time was too long, the melt could cool too much before the mold was filled. In evaporative pattern casting, the pouring time also controls the foam decomposition rate and gas evacuation, so the optimal pouring time would need to be determined separately. My orthogonal experiment provides a template for such optimization.
The temperature field simulation showed that the last solidified regions were the top of the head and the belly of the uncoupling cylinder. These regions coincided with the predicted shrinkage porosity. Because these regions were not in the main load path, the small amount of porosity was acceptable. If the porosity had appeared in the knuckle or shoulder, the coupler would have been rejected. The gating system and riser design successfully moved the shrinkage defects to non-critical regions. The stress simulation showed that the maximum stress was about 470 MPa, which was below the yield strength of the as-cast material. The hot cracking index was highest at the abrupt cross-section changes, but the actual stress was not high enough to cause cracks. The X-ray inspection confirmed this prediction. In evaporative pattern casting, the hot cracking risk could be different because the foam pattern can insulate the melt and change the cooling rate. Therefore, a similar stress simulation should be performed for evaporative pattern casting if that process is used.
The microstructure simulation revealed a mixture of equiaxed and columnar grains. The surface fine grain zone was beneficial for fatigue resistance. The equiaxed grains in the interior were beneficial for toughness. The columnar grains near the wall were not excessively coarse, and the heat treatment refined the microstructure further. The grain size after heat treatment was less than 1 μm in the tempered sorbite. This fine microstructure contributed to the high tensile strength and good elongation. The simulation also showed that the shoulder and knuckle regions had more equiaxed grains than the thinner sections. This was because the shoulder and knuckle were thicker and cooled more slowly, allowing more equiaxed nucleation and growth. The thinner sections cooled faster and formed more columnar grains. This result is consistent with classical solidification theory. In evaporative pattern casting, the foam pattern can also influence the local cooling rate, so the grain structure could be different. However, the CAFE model used here could be adapted to evaporative pattern casting by changing the heat transfer boundary conditions.
I also evaluated the filling process. The top-pouring system with side branches allowed the melt to fill the mold smoothly. The open riser at the top of the head provided a hot spot for feeding. The gating system was designed to avoid direct impingement on the mold walls and to reduce turbulence. The pouring time of 29 s was long enough to avoid excessive turbulence but short enough to prevent premature solidification. The shell preheating temperature of 425 °C reduced the thermal gradient between the melt and the mold, which helped to prevent cold shuts and misruns. The pouring temperature of 1 570 °C provided enough superheat to fill the thin sections. These parameters were validated by the simulation and the trial castings. In evaporative pattern casting, the filling process is more complex because the foam pattern decomposes and gas must escape through the coating. Therefore, the gating system for evaporative pattern casting would need to be designed differently, but the same principles of smooth filling and proper feeding would apply.
The heat treatment was also critical. The normalizing step homogenized the microstructure. The austenitizing at 910 °C for 2 h dissolved carbides and produced a uniform austenite. The quenching produced martensite. The tempering at 590 °C for 3 h produced tempered sorbite. The tempered sorbite had fine carbides distributed in a ferrite matrix. This structure provided high strength and good ductility. The hardness of 315 HBW was within the standard range. The yield strength of 810 MPa and tensile strength of 1 020 MPa exceeded the standard requirements. The elongation of 14.5% and reduction of area of 34.5% also exceeded the standard. These results confirmed that the combination of optimized investment casting and heat treatment produced a high-quality coupler. In evaporative pattern casting, similar heat treatment would be necessary, but the as-cast microstructure might be coarser, so the heat treatment parameters might need adjustment.
The X-ray inspection was performed on the head, knuckle connection, and base. No cracks or pores were detected. The metallographic examination showed that the microstructure was uniform and fine. The ferrite and pearlite in the as-cast condition were evenly distributed. After heat treatment, the tempered sorbite was fine and uniform. The energy-dispersive X-ray spectroscopy showed that carbon and manganese were uniformly distributed, and no large segregation was present. These results confirmed the reliability of the optimized process. I also compared the results with typical evaporative pattern casting results. In evaporative pattern casting, gas porosity and carbon pick-up can occur if the foam decomposition products are not properly removed. In investment casting, these problems are avoided because the pattern is removed before pouring. This is a significant advantage for safety-critical components such as locomotive couplers.
In summary, I optimized the investment casting process for a steel locomotive coupler using an orthogonal experiment and numerical simulation. 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. The temperature field simulation showed a favorable solidification sequence. The shrinkage porosity was confined to non-critical regions. The stress field simulation showed that the critical stress was below the material strength. The microstructure simulation showed a fine and uniform grain structure. The trial castings passed X-ray inspection and had mechanical properties above the standard. The heat-treated tensile strength reached 1 020 MPa, and the elongation reached 14.5%. These results demonstrate that the optimized investment casting process is suitable for producing high-integrity locomotive couplers. The same methodology can be adapted to evaporative pattern casting for similar components, although the thermal boundary conditions and gas evolution must be considered separately.
I also developed several quantitative relationships that can be used for future process design. The shrinkage porosity volume VSP can be expressed as a function of the process parameters:
$$
V_{SP} = f(T_p, T_s, t_p)
$$
where Tp is pouring temperature, Ts is shell preheating temperature, and tp is pouring time. The critical stress σ can be expressed as
$$
\sigma = g(T_p, T_s, t_p)
$$
These functions were evaluated numerically through the orthogonal experiment. The signal-to-noise ratio for the smaller-the-better criterion was calculated as
$$
S/N = -10 \log_{10}\left(\frac{1}{n}\sum_{i=1}^{n} y_i^2\right)
$$
where yi is the response value and n is the number of repetitions. In my experiment, each run was a single simulation, so n = 1. The S/N values were used to rank the factors. The range analysis was calculated as
$$
R = \max(\bar{y}_1, \bar{y}_2, \bar{y}_3, \bar{y}_4) – \min(\bar{y}_1, \bar{y}_2, \bar{y}_3, \bar{y}_4)
$$
where R is the range and \bar{y}j is the average response at level j. The factor with the largest range had the strongest influence. For VSP, the ranges were A > C > B. For σ, the ranges were also A > C > B. This confirmed that pouring temperature was the most important parameter. I also used the analysis of variance to determine the statistical significance. The P-value for factor A was less than 0.05, indicating a significant effect. The P-values for factors B and C were larger than 0.05, indicating smaller effects. These statistical results supported the optimization.
The thermal model also included the heat transfer coefficients. The heat flux between the casting and the shell was modeled as
$$
q = h_1 (T_c – T_s)
$$
where q is heat flux, h1 is the shell–casting heat transfer coefficient, Tc is casting temperature, and Ts is shell temperature. The heat flux between the shell and the environment was modeled as
$$
q = h_2 (T_s – T_0)
$$
where h2 is the shell–air heat transfer coefficient and T0 is ambient temperature. These equations were solved with the energy conservation equation:
$$
\rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + L \frac{\partial f_s}{\partial t}
$$
where ρ is density, Cp is specific heat, k is thermal conductivity, L is latent heat, and fs is solid fraction. This equation governed the temperature field and the solidification progress. The same equation would apply to evaporative pattern casting, but the boundary conditions would include the foam decomposition and gas gap. I did not include those effects in the present investment casting model.
The microstructure model also used the solid fraction and undercooling. The nucleation rate was integrated over the undercooling range. The grain density was calculated from the Gaussian distribution. The dendrite tip growth velocity was calculated from the KGT model. The KGT model can be written as
$$
V = \frac{D_1}{2\pi^2 \Gamma m (k-1) C_0} \left(\frac{\Delta T}{1 – \frac{2k}{…}}\right)^2
$$
However, the exact form of the KGT model depends on the alloy system and the assumed dendrite tip shape. In my simulation, I used the built-in KGT model in ProCAST, which uses the kinetic parameters listed in Table 3. The model predicted the grain structure reasonably well. The simulated microstructure was consistent with the experimental observations. The as-cast microstructure had ferrite and pearlite, and the heat-treated microstructure had tempered sorbite. The grain size was fine, and the mechanical properties were high. In evaporative pattern casting, the microstructure model would need to account for the foam decomposition products, which could affect the local carbon content and cooling rate. Therefore, evaporative pattern casting may require additional calibration of the CAFE model.
I also considered the economic and practical aspects. Investment casting requires a wax pattern, a ceramic shell, and a dewaxing step. Evaporative pattern casting requires a foam pattern, a refractory coating, and a sand backing. Investment casting provides better surface finish and dimensional accuracy, which are important for the coupler’s complex geometry. Evaporative pattern casting can reduce tooling costs for large parts, but it may produce more gas defects. For the locomotive coupler, the safety requirements favored investment casting. However, I believe that the orthogonal optimization method and the simulation approach developed here can be applied to evaporative pattern casting as well. In evaporative pattern casting, the key parameters would include foam density, coating permeability, sand compaction, pouring temperature, and pouring time. A similar L16 or L18 orthogonal array could be used to optimize those parameters.
The final optimized process was validated by trial production. The coupler castings had smooth surfaces and no visible defects. The X-ray inspection showed no internal cracks or pores. The metallographic examination showed a fine and uniform microstructure. The mechanical properties exceeded the standard requirements. The average tensile strength after heat treatment was 1 020 MPa, the yield strength was 810 MPa, the elongation was 14.5%, and the reduction of area was 34.5%. The hardness was 315 HBW. These values are suitable for heavy-haul railway service. The process is now ready for production. I also recommended that future work should include fatigue testing and low-temperature impact testing to fully qualify the coupler for service. In addition, a comparative study between investment casting and evaporative pattern casting for the same coupler geometry would be valuable to quantify the differences in microstructure and properties.
In conclusion, I successfully optimized the investment casting process for a steel locomotive coupler. The optimal parameters were pouring temperature 1 570 °C, shell preheating temperature 425 °C, and pouring time 29 s. The simulation results showed a favorable temperature field, low shrinkage porosity, acceptable stress levels, and a fine grain structure. The experimental results confirmed the simulation. The heat-treated coupler had a tensile strength of 1 020 MPa, which is above the standard requirement. The process is reliable and suitable for production. The methodology can also be adapted to evaporative pattern casting for similar steel components, although the differences in pattern decomposition and gas evolution must be considered. I believe that the combination of orthogonal experiment, casting simulation, stress analysis, and microstructure modeling is a powerful approach for developing high-integrity castings for railway applications.
| Process | Pattern material | Mold material | Pattern removal | Gas defects | Surface finish | Typical use |
|---|---|---|---|---|---|---|
| Investment casting | Wax or plastic | Ceramic shell | Dewaxing before pouring | Low | Excellent | Safety-critical precision parts |
| Evaporative pattern casting | Foam | Sand with refractory coating | Vaporized during pouring | Moderate to high | Good | Large complex parts |
The table above summarizes the main differences between investment casting and evaporative pattern casting. I used this comparison to justify my process selection. The coupler required excellent surface finish, dimensional accuracy, and low gas porosity. Investment casting met these requirements. Evaporative pattern casting remains a viable alternative for less critical parts or when tooling cost is the primary concern. However, for this locomotive coupler, investment casting was the better choice.
I also derived an empirical relationship between pouring temperature and shrinkage porosity based on the orthogonal experiment. The trend showed that increasing the pouring temperature from 1 530 °C to 1 570 °C reduced the shrinkage porosity, but increasing it further to 1 590 °C increased the porosity. This indicated an optimal pouring temperature. The shell preheating temperature had a smaller effect, but a higher shell temperature reduced the thermal gradient and improved feeding. The pouring time had a moderate effect. A shorter pouring time increased the filling velocity, which could cause turbulence, while a longer pouring time increased the risk of premature solidification. The optimal pouring time was 29 s. These trends are consistent with foundry practice. In evaporative pattern casting, the pouring temperature would also need to be optimized because too low a temperature would not vaporize the foam completely, and too high a temperature could cause excessive gas generation.
The stress analysis showed that the maximum stress occurred at the bottom of the coupler during solidification. The stress was about 470 MPa at 6 000 s. This stress was below the yield strength of the as-cast material, which was not measured but can be estimated from the tensile strength. The hot cracking index was highest at the abrupt section changes. The shoulder and base junctions had higher hot tearing susceptibility. However, the actual stress was not high enough to cause cracks. The X-ray inspection confirmed this. The stress simulation also showed that the node at the shoulder gate had lower stress because the geometry allowed free contraction. This result highlights the importance of gating design in reducing hot tearing. In evaporative pattern casting, the gating design would also need to minimize hot spots and allow free contraction.
I also analyzed the effect of heat treatment on the microstructure and properties. The as-cast microstructure was ferrite and pearlite. The ferrite grains were coarse, with a maximum length of about 100 μm. The tensile strength was only 625 MPa, and the elongation was 2.7%. After heat treatment, the microstructure became tempered sorbite with a grain size less than 1 μm. The tensile strength increased to 1 020 MPa, and the elongation increased to 14.5%. The hardness increased from 238 HBW to 315 HBW. These changes are typical for low-alloy cast steel. The heat treatment refined the microstructure and improved both strength and ductility. In evaporative pattern casting, the as-cast microstructure might be coarser due to slower cooling, so the heat treatment might need to be adjusted to achieve the same properties. This is another reason I selected investment casting.
The energy-dispersive X-ray spectroscopy showed that carbon and manganese were uniformly distributed. No large segregation was observed. This is important because segregation can cause local variations in hardness and toughness. The uniform distribution was achieved through proper melting, pouring, and solidification control. The optimized pouring temperature and shell preheating temperature helped to reduce segregation. In evaporative pattern casting, segregation can also occur if the foam decomposition products affect the local carbon content. Therefore, evaporative pattern casting may require additional process control to avoid carbon pick-up or carbon loss.
Overall, my study demonstrated that the optimized investment casting process produced a high-quality locomotive coupler. The process parameters were determined by orthogonal experiment and validated by simulation and experiment. The coupler had a fine microstructure, no internal defects, and excellent mechanical properties. The tensile strength of 1 020 MPa exceeded the standard requirement of 830 MPa. The elongation of 14.5% exceeded the standard requirement of 14%. The hardness of 315 HBW was within the standard range of 241–311 HBW. The yield strength of 810 MPa exceeded the standard requirement of 590 MPa. The reduction of area of 34.5% exceeded the standard requirement of 30%. These results confirm that the process is suitable for production. I also believe that the methodology can be applied to evaporative pattern casting and other precision casting processes. The key is to identify the critical process parameters and use orthogonal experiments and numerical simulation to optimize them.
Future work should include a full-scale fatigue test and a low-temperature impact test to qualify the coupler for heavy-haul service. It would also be useful to compare the fatigue life of investment cast couplers with evaporative pattern casting couplers. The microstructure simulation could be extended to predict fatigue crack initiation. The stress simulation could be coupled with service loads to predict fatigue life. These extensions would provide a more complete understanding of the coupler performance. However, the current study provides a solid foundation for process optimization and quality control.
I also note that the orthogonal experiment used a limited number of levels and runs. A more detailed optimization could use a response surface methodology or a genetic algorithm. However, the L16 orthogonal array was sufficient to identify the significant factors and the optimal range. The pouring temperature was the most important factor. The shell preheating temperature and pouring time had smaller effects. The optimal parameters were robust because the responses were not highly sensitive to small changes in shell temperature and pouring time. This is beneficial for production because it provides a wider processing window. In evaporative pattern casting, the processing window might be narrower because of the foam decomposition. Therefore, a similar robustness study would be valuable for evaporative pattern casting.
In summary, I optimized the investment casting process for a steel locomotive coupler. The optimal parameters were pouring temperature 1 570 °C, shell preheating temperature 425 °C, and pouring time 29 s. The simulation showed a good temperature field, low shrinkage porosity, acceptable stress, and a fine microstructure. The experimental results confirmed the simulation. The heat-treated coupler had a tensile strength of 1 020 MPa and an elongation of 14.5%. The process is reliable and suitable for production. The methodology can be adapted to evaporative pattern casting and other precision casting processes. I believe that this work provides a useful reference for the production of high-integrity steel castings for railway applications.