As a critical component of railway infrastructure, the crossing determines the safety and efficiency of train operations. The failure of crossing components, particularly in the heartrail and wingrail regions, accounts for over 50% of all track-related failures. Through my research, I identified that the primary cause of premature crossing failure is not material fatigue but rather the presence of casting defects introduced during the manufacturing process. The necessity to improve casting quality through systematic investigation of process parameters cannot be overstated. This research addresses the challenge of reducing shrinkage porosity and cavity defects in sand casting parts through comprehensive numerical simulation and process optimization. The high manganese steel used in these sand casting parts exhibits exceptional work-hardening capability and impact toughness, yet its poor castability, characterized by a solidification shrinkage rate exceeding 8.5%, necessitates meticulous process control to produce sound sand casting parts.
To investigate the mechanisms governing defect formation in high manganese steel crossings, I conducted a comprehensive numerical simulation study using ProCAST finite element software. This investigation establishes mathematical models and geometric representations for monoblock crossings, incorporating the nonlinear thermophysical properties of the high manganese steel and simulating the complete filling and solidification process for typical sand casting parts in railway applications.
Establishment of the Research Model and Mathematical Foundations
The geometric model of the crossing system forms the foundation for accurate numerical simulation of sand casting parts. The crossing geometry was established using SolidWorks, capturing all critical features including the heartrail, wingrail, flanges, and wheel contact surfaces. The casting length of the crossing under investigation is 4,100 mm with the complete mold assembly extending to 7,100 mm. This substantial size presents considerable challenges in mesh generation and computational efficiency. A differential meshing strategy was adopted, where the critical heartrail and wingrail regions received fine mesh elements of 5 mm size to capture local heat transfer phenomena accurately, while the mold regions, which require less resolution, utilized 100 mm elements. The final mesh contained 3,657,221 volume elements, with approximately 1.5 million elements concentrated in the regions of greatest interest.
The riser system was designed based on the modulus method, incorporating the geometric characteristics of the casting. The dimension parameters of the risers are summarized in the following table, which is essential for understanding the feeding behavior of these large sand casting parts:
| Parameter | Symbol | Value |
|---|---|---|
| Metal shrinkage rate | ε (%) | 8.5 |
| Riser modulus | RM (cm) | 4.33 |
| Riser top diameter | UD (cm) | 132 |
| Riser bottom diameter | 0D (cm) | 222 |
| Riser height | H (cm) | 222 |
The governing equations for the casting process simulation through sand casting parts manufacturing include the continuity equation and momentum equations that account for the incompressible Newtonian fluid behavior of the molten metal. The continuity equation expresses the conservation of mass, where the density of the molten metal varies nonlinearly between the liquidus and solidus temperatures. The momentum equations, expressed in three orthogonal directions, incorporate gravitational forces and viscous effects. To accurately describe the thermal behavior, the energy equation couples the temperature field with fluid flow, incorporating the latent heat of fusion through the enthalpy method. These governing equations are completed by the Niyama criterion for defect prediction, which provides a dimensionless parameter that correlates with microporosity formation:
$$Ny = \frac{G}{\sqrt{R}}$$
where G represents the temperature gradient (K/cm) and R is the cooling rate (K/min). Regions where Ny falls below the critical threshold of 0.85 (K·s)1/2/mm are considered susceptible to shrinkage porosity formation.
The high manganese steel used in this research has the chemical composition shown in the subsequent table, which was utilized to calculate the thermophysical properties as temperature-dependent functions. This non-linear treatment is essential for accurately modeling the solidification behavior of sand casting parts in actual production scenarios.
| Element | C | Mn | Si | P | S |
|---|---|---|---|---|---|
| Weight percentage (%) | 1.1 | 13.0 | 0.53 | 0.033 | 0.008 |
One significant challenge in simulating the inclined casting process for these sand casting parts arises from the overflow of molten metal from the riser during filling. In the actual production of sand casting parts, a constant pouring rate of 50 kg/s is maintained, with the molten metal flowing through the pouring cup into the sprue then through the runner system. However, for the inclined pouring configuration, metal overflow from the risers causes the net flow rate into the mold cavity to decrease over time. This complexity creates a significant challenge for standard numerical simulation approaches, which typically cannot handle such dynamic boundary conditions effectively.
To address this issue, I innovatively developed a pouring velocity decay assumption that modifies the inlet velocity boundary condition. The metal overflow from the riser is equivalent to a pressure relief at the riser top, allowing air ingress but preventing metal escape in the simulation model. This equivalent pressure modeling approach maintains consistency with actual pressure distribution in the casting system. The actual metal loss through riser overflow was determined through multiple simulation iterations. Testing revealed that approximately 21.1% of the total molten metal mass exits through the riser tops during the filling process for typical sand casting parts.
The pouring velocity decay curve was determined by solving a system of nonlinear equations that couple the total casting time, the cumulative metal mass entering the mold, and the boundary conditions. The exponential decay function was selected based on the physical behavior of the system, where each riser encountered by the molten metal front contributes to the reduction in effective flow rate:
$$u_1 = 52.07 – 2.07e^{\frac{t-t_0}{0.12}}$$
For subsequent risers, a damping factor was introduced to account for the decreasing overflow at each successive riser position. This iteration continued until all eight risers were included, producing the final pouring velocity curve that drives the filling simulation. The mathematical treatment of this phenomenon demonstrates how sophisticated modeling techniques can capture the complex reality of producing large sand casting parts for railway applications.
Casting Process Visualization and Defect Analysis
The casting process for high manganese steel crossing was visualized through comprehensive numerical simulation, allowing examination of the temperature field, fluid flow patterns, and defect formation mechanisms. The pouring temperature was set at 1,450°C with the liquidus and solidus temperatures calculated as 1,392°C and 1,019°C respectively. The complete filling of these sand casting parts required 22.56 seconds, during which the flow field remained relatively stable throughout the mold cavity. Minor flow disturbances were observed only near the pouring end where the molten metal enters the casting system.
The casting process was separated into five distinct, though overlapping, stages for detailed analysis: the filling stage, the exothermic reaction stage, the solidification stage, the feeding stage, and the defect formation stage. These stages interact continuously in time and space, creating complex thermomechanical behavior within the sand casting parts. In the filling stage, molten metal enters the mold and sequentially ignites the exothermic risers. The early contact between molten metal and the external chills creates a thin equiaxed crystal layer at these surfaces, establishing the foundation for directional solidification that controls the solidification pattern in these sand casting parts. The exothermic risers, containing aluminum thermite, fluxing agents, and oxidizers, sustain high temperatures within the riser cavity, significantly reducing the cooling rate of the riser metal and extending the effective feeding time. This extension is critical for producing sound sand casting parts that can withstand the demanding service conditions of railway crossings.
As solidification proceeds, the temperature gradients establish the directional solidification pattern that governs feeding efficiency. The mold was configured with the heartrail and wingrail surfaces facing downward, allowing these critical regions to solidify first and be feeding more effectively. The chill blocks placed in the mold accelerate local solidification at the working surfaces, while the risers maintain liquid metal at elevated temperatures to facilitate flow into solidifying regions. This coordinated chill-riser system directs the solidification front in a manner that progressively moves defects toward the riser area, away from critical working surfaces in these large sand casting parts. The temperature monitoring showed that the peak cooling rate occurs at the chill contact points, which cool rapidly to the solidus temperature, while the riser regions exhibit delayed cooling commensurate with their role as feeders.
Analysis of the temperature fields across the casting cross-section reveals important relationships between cooling patterns and defect formation. During the late stages of solidification, the interior regions approach the liquidus temperature while maintaining sufficient fluidity for effective feeding. The solidification time maps show that complete solidification of these sand casting parts requires approximately 3,000 seconds, with the regions adjacent to the chills solidifying first, followed by the progressive advancement of the solidification front toward the risers. The formation of shrinkage defects follows a predictable pattern based on the temperature gradient and cooling rate distribution. Defects tend to concentrate in regions where the Niyama criterion falls below the critical value, creating localized areas of microporosity that can initiate fatigue cracks during railway service.
Through analysis of defect distribution in the simulation results, I found that each riser creates a characteristic “V-shaped” feeding zone beneath it, extending a distance proportional to the spacing between adjacent risers. Between adjacent risers, arcuate regions form where feeding is insufficient, leading to defect concentration. The edge risers exhibit more severe defect accumulation on their outer boundaries where the feeding distance becomes excessive for the local casting geometry. These findings provide critical insight into the feeding limitations of sand casting parts and guide the optimization of riser placement for improved casting quality.
Analysis of Different Exothermic Riser Parameters
In investigating the influence of riser parameters on casting quality, I systematically compared the effects of different exothermic riser configurations. The risers were characterized by their heat release energy, burning time, and ignition temperature with the dimensions, as provided in the table below. The standard insulating risers, without exothermic capability, provide only insulation to slow cooling, while exothermic insulating risers actively generate heat through chemical reactions to maintain the riser metal at elevated temperatures.
| Riser Type | Exothermic Energy (kJ/kg) | Burning Time (s) | Ignition Temperature (°C) | Shrinkage Porosity (%) | Shrinkage Cavity (%) |
|---|---|---|---|---|---|
| Insulating | 0 | 0 | 0 | 23.10 | 1.002 |
| Exothermic 1 | 400 | 240 | 1,000 | 19.69 | 0.670 |
| Exothermic 2 | 1,200 | 240 | 1,000 | 16.51 | 0.514 |
| Exothermic 3 | 2,000 | 240 | 1,000 | 15.84 | 0.499 |
| Exothermic 4 | 1,200 | 120 | 1,000 | 17.01 | 0.530 |
| Exothermic 5 | 1,200 | 360 | 1,000 | 16.17 | 0.504 |
The investigation of riser burning time revealed that for these large sand casting parts, variations in burning time within the range of 120 to 360 seconds produced minimal differences in casting quality. The shrinkage porosity varied from 17.01% at 120 seconds to 16.17% at 360 seconds, representing a difference of less than 5%. This relative insensitivity to burning time arises from the fact that the total cooling time for these sand casting parts exceeds 6,000 seconds, making the exothermic burning duration a small fraction of the total solidification time. Consequently, the additional heat input from extended burning, while beneficial in maintaining riser temperature, does not significantly alter the solidification pattern or feeding efficiency in such large castings.
In contrast, the exothermic energy content of the risers had a substantial influence on the quality of sand casting parts. Increasing the exothermic energy from 0 to 400 kJ/kg reduced shrinkage porosity from 23.10% to 19.69%, while increasing to 1,200 kJ/kg further reduced porosity to 16.51%. Beyond this value, the improvement became marginal, with 2,000 kJ/kg achieving only 15.84% porosity. The same trend appears in the shrinkage cavity fraction, which decreases from 1.002% for insulating risers to 0.514% at 1,200 kJ/kg. The results suggest that 1,200 kJ/kg represents the practical upper limit for effective feeding in this casting configuration. Higher heat outputs provide negligible additional benefit while potentially introducing adverse effects such as localized thermal gradients that delay mold opening, increase energy consumption, and may enhance the risk of hot tearing in the cast structure.
The mechanism by which exothermic risers improve casting quality lies in their ability to extend the duration of effective liquid feeding. The exothermic reaction maintains the riser metal at temperatures well above the liquidus for extended periods after the onset of solidification in the casting body. This sustained high temperature drives fluid flow through the feeding channels, compensating for solidification shrinkage and maintaining continuity in the liquid metal. During the critical period when the casting interior is passing through the mushy zone, the riser metal retains sufficient fluidity to respond to the pressure differentials created by solidification contraction, thereby minimizing microporosity formation in the sand casting parts.
Solidification and Feeding Characteristics
The solidification behavior of the high manganese steel crossing was examined through thermal analysis at specific monitoring points strategically positioned within the casting. The temperature histories at these points reveal distinct cooling regimes corresponding to different regions within the sand casting parts. The chill contact points exhibit an initial rapid temperature drop as the molten metal contacts the relatively cool chill blocks. Within seconds, the metal temperature approaches the solidus, initiating solidification at this location. Interestingly, during the 1,000 to 2,000 second interval, these points experience a slight temperature increase, indicating remelting of the initially formed solidification layer. This remelting phenomenon occurs because the relatively thin solidified shell, once formed, provides insulation that reduces the heat extraction rate, while the interior metal continues to supply heat through conduction and convection.
The core regions of the crossing show a more gradual temperature evolution. After pouring, the core temperature drops through the liquidus and then proceeds more slowly through the mushy zone as it approaches complete solidification. The temperature during the critical period between 1,000 and 2,500 seconds remains above the solidus, allowing fluid flow and feeding to continue. This extended period of adequate fluidity is essential for effective feeding of the solidifying structure, as it allows molten metal to flow from the riser to compensate for solidification shrinkage at the solidification front.
In contrast, the riser regions do not exhibit any rapid temperature drops after pouring. The exothermic reaction sustains the riser metal temperature, creating a flat-bottomed temperature profile that maintains elevated temperatures for extended periods. The exothermic riser temperature curves show a characteristic upward convex shape, indicating that the internal heat generation initially exceeds the heat losses to the surroundings. Only after the exothermic reaction subsides does the cooling rate increase, and the riser metal begins to solidify. By comparison, the insulating riser without exothermic capability shows only delayed cooling without the temperature maintenance effect, leading to reduced feeding efficiency.
According to the Niyama criterion for defect prediction in sand casting parts, values below the critical threshold indicate regions with significant microporosity formation tendency. The temperature data were used to calculate the Niyama parameter distribution throughout the casting. This information helped clarify why the exothermic riser at 1,200 kJ/kg provides optimal performance for these sand casting parts. At this heat output level, the riser maintains the feeding capability precisely through the critical solidification window of the crossing, maximizing the effective feeding distance without extending the total solidification time excessively. This balance is reflected in the defect distributions, which show defects concentrated near the riser area and around bolt holes, but essentially no defects in the heartrail and wingrail regions of the sand casting parts.
The finite element simulation also reveals an important phenomenon at the chill locations: the differential cooling rates can actually increase rather than decrease defect formation tendencies in certain conditions. While chills enhance local cooling rate, creating fine-grained structures at the casting surface, they also establish steep temperature gradients that increase the pressure drop needed to maintain feeding flow. In the absence of adequate riser capacity at these locations, the positive effects of grain refinement are counteracted by increased susceptibility to microporosity. This finding highlights the importance of coordinated design of chill placement and riser capacity in sand casting parts to achieve balanced solidification control.
Riser Optimization for Large Sand Casting Parts
Based on the investigation of the relationship between solidification patterns and defect formation in large sand casting parts, we identified a critical transition region between the pouring-end open riser and the nearest blind riser. This region, spanning 800 mm, lacked any dedicated feeding module in the original casting design. The maximum distance between adjacent riser feeding zones created a substantial arcuate region of inadequate feeding, characterized by severe shrinkage defects. The local porosity in this region reached 36.17%, substantially degrading the structural integrity of these sand casting parts.
To address this defect concentration, rather than attempting to modify the existing riser parameters, which the preceding analysis demonstrated would be insufficient, we designed a riser optimization scheme that added a new riser at the transition region. The placement of this new riser was determined based on the “V-shaped” feeding zone prediction derived from the earlier analysis. This prediction tool proved effective for establishing the processing parameters of sand casting parts.
The numerical simulation of the modified casting process demonstrated substantial improvement in the local casting quality. The local shrinkage porosity decreased from 36.17% to 22.11%, while the local shrinkage cavity fraction decreased from 1.13% to 0.69%. The predicted defect distribution at the riser feeding boundaries closely matched the simulation results, validating the predictive capability of the V-zone approach. Notably, we observed that the defects originally concentrated at the bolt hole region migrated upward, concentrating above the bolt holes rather than below. This redistribution occurred because the wall thickness at that location is different, with the track region having a thickness h that is 80 mm thinner than the surrounding regions in the sand casting parts. The thinner section solidifies earlier, creating a solidified barrier that prevents the upward flow of feeding metal to the bolt hole region, while the thicker section maintains fluidity longer, allowing more effective feeding below the bolt holes. This phenomenon confirmed the earlier hypothesis that defects in the bolt hole region of the sand casting parts could not be eliminated by adjusting the riser parameters alone.
Furthermore, this upward migration of defects indicates that the metal density in the region near the working surface, which is located at the bottom of the casting during the pouring orientation, is improved. The near-surface metal densities achieved correspond to significantly enhanced mechanical properties. This optimized configuration of the sand casting parts achieved total tonnage capacities of up to 300 million tons, substantially exceeding the capacity of conventional designs. The results demonstrate that understanding the relationship between solidification modes and defect formation enables systematic improvement of sand casting parts for demanding railway applications.
A Novel Forging-Casting Integrated Production Method for High Manganese Steel Crossings
The failure of crossings in railway service occurs primarily through the heartrail and wingrail regions, with defects such as spalling, impact wear, crushing, and horizontal cracking accounting for more than 80% of all failure modes. According to the mechanism investigation on defect formation in these regions, conventional approaches using external chills and riser systems cannot entirely eliminate internal defects. This motivated the innovation of an integrated forging-casting production method that creates hybrid metallic structures within the casting, replacing conventional casting entirely in the most critical zones.
The fundamental concept behind this innovation is to manufacture the highly stressed heartrail and wingrail components as forged preforms, which are then incorporated into the sand mold as internal chills. These forged components, produced through hot forging and machining, exhibit superior mechanical properties compared to their cast counterparts. The forged pieces are pre-processed with connection slots on the interface surface to enhance bonding with the cast metal during pouring. The assembly of these preforms within the mold cavity positions them exactly at the regions where defects most severely compromise casting quality, effectively replacing the weakest portions of the casting with high-quality wrought material.
During pouring, the molten metal surrounds these forged preforms and initiates localized melting at the interface. The thermal interaction between the molten metal and the preform surface occurs in four sequential stages. In the first stage, the molten metal contacts the relatively cool forged surface and rapidly cools, creating a thin layer of fine equiaxed crystals at the interface. These fine grains provide a favorable nucleation substrate for subsequent solidification. In the second stage, the heat from the molten metal begins to raise the preform surface temperature toward the solidus, while the chilled region around the preform continues solidifying at a moderate rate. The initial fine equiaxed structures provide a foundation for the subsequent columnar growth.
In the third stage, the preform surface temperature rises above the solidus, causing remelting of the initially solidified metal near the interface. This remelting creates favorable conditions for metallurgical bonding between the forged preform and the cast metal by allowing the two materials to intermix at the atomic level. In the fourth stage, as both materials cool together, solidification occurs continuously across the interface, creating a seamless metallic bond. Successful execution of this process requires careful control of the preform temperature, mass, and pouring parameters to ensure that sufficient melting occurs at the interface while preventing complete fusion or erosion of the preform structure.
The optimal configuration involves a three-piece separated preform design: one heartrail component and two wingrail components, preheated to 400°C before pouring. The numerical simulation results showed that this configuration achieves the most favorable temperature distribution at the forging-casting interface, sufficient to facilitate metallurgical bonding without excessive degradation of the forged microstructure. Preheating the preform is essential for achieving the required interfacial temperature rise. Without preheating, the thick forged sections of the sand casting parts do not reach melting temperature even at their outer surfaces, except in regions immediately below the risers. Preheating to 800°C provides abundant melting but risks degradation of the forged microstructure and excessive oxidation. The 400°C preheating condition emerges as the optimal balance, providing adequate interfacial bonding while maintaining the integrity of the forged material.
CAFE Modeling for Microstructure Evolution
To further understand the solidification process of high manganese steel crossing sand casting parts, I employed the CAFE (Cellular Automaton Finite Element) model within the ProCAST software. This coupled model provides a powerful tool for predicting the evolution of grain structure during solidification, which is critical for optimizing the mechanical properties of sand casting parts used in demanding railway applications.
The CAFE model combines finite element heat transfer analysis with cellular automaton grain growth simulation. The finite element mesh computes the macroscopic temperature field, while the cellular automaton grid, superimposed on the finite element mesh, tracks the growth of individual grains based on local thermodynamic conditions. The coupling between these two scales occurs through temperature interpolation at each cellular automaton cell, using the nodal temperatures from the finite element solution. This interpolation procedure enables grains to sense the local thermal environment and grow accordingly.
The nucleation model employed is based on a Gaussian distribution of nucleation sites, as expressed by the continuous non-uniform nucleation equation:
$$\frac{dn}{d(\Delta T)} = \frac{n_{max}}{\sqrt{2\pi}\Delta T_\sigma} \exp\left[-\frac{1}{2}\left(\frac{\Delta T – \Delta T_n}{\Delta T_\sigma}\right)^2\right]$$
where nmax is the maximum nucleation density, ΔTσ is the standard deviation of the nucleation undercooling, ΔTn is the mean nucleation undercooling, and ΔT is the local undercooling. The grain growth kinetics are described by the Lipton-Glicksman-Kurz dendrite tip growth model, accounting for the thermal and solutal undercooling contributions:
$$v = \alpha_2 \Delta T^2 + \alpha_3 \Delta T^3$$
where the growth coefficients α₂ and α₃ are determined from the material properties. For the high manganese steel in this study, the growth coefficients were calculated as α₂ = 5.6×10⁻⁷ m·s⁻¹·K⁻² and α₃ = 2.8×10⁻⁷ m·s⁻¹·K⁻³. These coefficients quantify the relationship between local undercooling and dendrite tip velocity for sand casting parts manufactured from this alloy.
The simulation parameters for the CAFE model were determined based on both theoretical considerations and empirical calibration against observed microstructures of sand casting parts. The surface nucleation density was set to 1.7×10⁵ m⁻², representing the number of potential nucleation sites at the mold surface. The volume nucleation density was 5.1×10⁷ m⁻³, representing the nucleation sites within the bulk of the molten metal. The mean surface nucleation undercooling was 1 K with a standard deviation of 0.1 K, while the mean volume nucleation undercooling was 10 K with a standard deviation of 0.1 K. These parameters were adjusted through iterative comparison between simulation results and microstructural observations from actual production of sand casting parts.
The CAFE simulation results reveal the evolution of the solidification microstructure during crossing production. In the initial stage, immediately after pouring, a thin layer of fine equiaxed grains forms at the mold surface. This layer, sometimes called the chill zone, arises from the severe undercooling experienced by the metal at the mold interface, which activates a high density of surface nucleation sites. However, this chill zone is extremely thin, often barely visible in the overall casting cross-section, since the thermal resistance at the mold-metal interface quickly reduces the heat flux, decreasing the undercooling and halting further nucleation.
Following the formation of the chill zone, columnar dendrites grow inward from the mold surface, oriented perpendicular to the mold wall. These columnar grains, which have their preferred crystallographic orientations roughly aligned with the heat flow direction, grow competitively, with those optimally oriented grains overgrowing their less favorably oriented neighbors. This competitive growth process produces the columnar zone characteristic of this region in sand casting parts.
As solidification progresses, the temperature gradient ahead of the columnar front decreases, and equiaxed grains begin to nucleate in the remaining liquid. These equiaxed grains continue to grow, eventually blocking the columnar front and forming an equiaxed zone at the center of the casting. This columnar-to-equiaxed transition (CET) is clearly visible in the simulated grain structure, and the transition position depends on the competition between the growth rate of columnar dendrites and the nucleation and growth of equiaxed grains in the constitutionally undercooled liquid ahead of the columnar front.
Validation of the CAFE simulation for these sand casting parts involved comparison with acid-etched macrostructures from actual production. The comparison showed excellent agreement in the grain structure characteristics: the thin surface chill zone, the intermediate columnar zone, and the central equiaxed zone all appeared in the correct proportions in the simulation. The grain sizes in the simulation matched the experimentally observed values, confirming that the Gaussian parameters for nucleation and growth were accurately calibrated.
With this validated CAFE model, I extended the analysis to examine the solidification behavior of the forged-casting integrated sand casting parts design. The simulation included the forged preform within the mold and tracked the grain structure evolution during solidification. The results showed that at the interface between the forged preform and the cast metal, the grain structure differs from that at the sand mold interface. The initial fine equiaxed grain layer present at the mold surface is absent at the forged surface, and instead, columnar grains grow directly from the interface. This occurs because the forged preform surface, being heated by the surrounding molten metal, does not provide the same high undercooling necessary for the mass nucleation of fine equiaxed grains at the mold surface.
When the interface temperature exceeds the solidus, the prior solidified layers near the forged preform surface remelt, initiating the metallurgical bonding process. During the subsequent solidification, columnar grains grow epitaxially from the partially melted preform surface, creating a firm connection between the forged and cast portions of the sand casting parts. The simulation results confirm that the separated forging-casting strategy produces sound bond between the metallic components, further validating the theoretical feasibility of the proposed approach.
Conclusions and Implications for Sand Casting Parts Manufacturing
This comprehensive investigation of casting process parameters for high manganese steel crossing addressed key challenges in the production of high-quality sand casting parts for demanding railway applications. Through systematic numerical simulation, process optimization, and theoretical analysis, several important conclusions emerge that have implications for the foundry practice of these components.
The casting process of high manganese steel crossings was fully visualized and separated into five distinct but interconnected stages: filling, exothermic reaction, solidification, feeding, and defect formation. These stages progress concurrently and influence each other continuously, creating a complex metallurgical environment within the mold. The analysis reveals that the control of solidification sequence through coordinated placement of chills and risers is the primary lever for minimizing casting defects in sand casting parts. The exothermic capacity of the riser system emerges as the dominant factor in determining casting quality, with 1,200 kJ/kg representing the optimal heat release value for this casting configuration. Beyond this level, further exothermic energy provides minimal improvement while increasing costs and potential quality risks. The burning time of the exothermic risers is of secondary importance for these large-scale sand casting parts, as the total solidification time significantly exceeds the duration of the exothermic reaction.
The riser optimization scheme, based on the V-shaped feeding zone prediction, successfully reduced shrinkage porosity from 36.17% to 22.11% in the critical transition region of the sand casting parts. This approach also demonstrated predictive capability for defect distribution, offering a design methodology that can extend to other sand casting parts with similar feeding configurations. The defects in the bolt hole regions cannot be eliminated solely through riser optimization, as they lie beyond the effective feeding distance of the riser system. Addressing these defects requires either the addition of supplementary feeding devices or changing the local geometry through design modification.
The innovative forging-casting integrated production method offers a breakthrough approach for enhancing the reliability and service life of crossings. By producing the most critical zones of the sand casting parts as forged preforms that are incorporated and metallurgically bonded during casting, this method effectively eliminates internal defects in regions prone to premature failure. The numerical simulation and CAFE analysis confirm the feasibility of achieving strong metallurgical bonding at the forging-cast interface when the preforms are preheated to 400°C, providing a new pathway for manufacturing high-performance sand casting parts for railway infrastructure.
The CAFE model developed and validated in this research provides a capable tool for predicting the solidification grain structure of sand casting parts. The agreement between simulation results and actual production confirms that the Gaussian nucleation parameters calibrated for the high manganese steel are reliable for predicting grain characteristics. This validated tool can be employed for future optimization of solidification microstructure in sand casting parts, potentially enabling the design of process parameters that produce tailored grain structures for specific performance requirements. The successful application of this modeling approach to the forging-casting integrated sand casting parts design demonstrates its utility in evaluating new manufacturing concepts before physical prototype trials.

