In modern railway and heavy machinery manufacturing, box-type steel castings are widely used as key structural components such as bearing housings, gearbox casings, bogie frames, and motor supports. These components often operate under severe cyclic loading and demand high reliability. However, their complex geometries, varying wall thicknesses, and the inherent solidification characteristics of steel make them prone to various casting defects, including shrinkage cavities, shrinkage porosity, hot tears, cold cracks, gas porosity, and sand inclusions. The presence of any casting defect not only reduces the load-bearing capacity of the component but also poses a serious threat to operational safety. Therefore, understanding the formability of box-type steel castings and systematically analyzing the formation mechanisms of casting defects are of great significance for process optimization and quality assurance.
Numerical simulation has emerged as a powerful tool to predict and eliminate casting defects. By solving the governing equations of fluid flow, heat transfer, and solidification, one can visualize the filling and solidification processes, identify hot spots, and assess the feeding efficiency of risers. In this study, I adopted a commercial finite-element package as the simulation platform, combined with a pre-processing module for three-dimensional solid modeling, to investigate the casting process of two representative box-type steel castings: a traction motor suspension housing (the so-called “axle box body”) and a bogie bracket for a high-power diesel locomotive. The aim was to predict casting defects such as shrinkage and hot tearing, optimize the gating and risering system, and finally verify the improved designs by actual production trials.
This article is organized as follows: First, the significance and development of casting process simulation are summarized. Second, the preparatory work for producing box-type steel castings, including electric arc furnace melting, sand mold materials, and common casting defects, is discussed. Third, the detailed case studies of two typical components are presented, with emphasis on the application of thermal and stress field simulation to resolve cracking and shrinkage problems. Finally, conclusions are drawn based on the successful production verification.
1. Introduction and Background
With the rapid development of railway transportation toward heavy freight and high-speed passenger services, locomotive components are subjected to increasingly demanding service conditions. Box-type castings, such as main bearing supports and axle box housings, are crucial for the running gear of locomotives. The quality of these castings directly affects the overall performance and safety of the vehicle. In the past, casting process design was mainly based on empirical rules and trial-and-error methods. However, the increasing complexity of product structures and the requirement for defect-free castings have necessitated a more scientific approach.
The application of computer simulation to casting processes allows engineers to numerically analyze the filling, solidification, and cooling stages before the actual pouring. This helps in predicting potential casting defects and optimizing process parameters accordingly. The benefits include reduced development time, lower prototyping costs, improved yield, and enhanced product quality. In this context, I have employed the finite-element simulation software to model the stress evolution and thermal fields during solidification, and the finite-volume-based flow solver to simulate mold filling and shrinkage porosity. These tools enabled me to identify the root causes of casting defects in the studied components and to evaluate multiple design alternatives virtually.
2. Numerical Simulation of Casting Processes
2.1 Significance of Casting Process Simulation
Casting solidification involves complex heat transfer, fluid flow, and phase transformation phenomena. The objective of numerical simulation is to obtain the transient distributions of temperature, velocity, and stress, and then to use appropriate criteria to predict defects such as shrinkage cavities, porosity, hot tears, and cold cracks. With the help of simulation, one can answer the following questions:
- Where will the last solidified regions be located?
- Will the risers be effective in feeding the liquid shrinkage?
- What is the optimal size and position of chills and insulating sleeves?
- What is the magnitude of residual stress and the risk of hot tearing?
- How can the gating system be designed to avoid entrainment of air and mold erosion?
In the present work, simulation was used to optimize the casting process of box-type components, particularly to eliminate shrinkage porosity and hot cracks. The simulation results were validated by sectioning and macroscopic examination of actual castings.
2.2 Development of Casting Simulation Technology
The research on solidification simulation began in the 1960s with the pioneering work of Forsund and others who used finite difference methods to solve one-dimensional heat conduction problems. Since then, the field has evolved rapidly. The early stage focused on temperature field calculations and simple shrinkage predictions. In the 1980s, flow field simulation was added, and in the 1990s, commercial software packages appeared, enabling industrial applications. More recently, microstructural simulation and multi-physics coupling have become active research topics. The current state-of-the-art simulation software can handle not only heat transfer and fluid flow but also stress analysis, deformation, and microstructural evolution.
For the simulation of casting defects, the key is to establish reliable criteria. Among them, the well-known \(Niyama\) criterion is widely used for shrinkage porosity prediction in steel castings. It is defined as
$$ G/\sqrt{R} = Niyama$$
where \(G\) is the local temperature gradient and \(R\) is the cooling rate. A low value of \(G/\sqrt{R}\) indicates insufficient feeding and a high risk of microporosity. Another important parameter is the critical solid fraction for liquid flow, \(f_{s,crit}\), typically around 0.7–0.8. When the local solid fraction exceeds this value, the liquid becomes immobile and feeding stops.
2.3 Simulation Platform and Workflow
In this study, I used a pre-processing software (the “CAD” module of a commercial simulation system) to create three-dimensional solid models of the castings, including gating systems, risers, and chills. The models were then transferred to the simulation platform where a finite-difference mesh was generated. The main steps are summarized below:
| Step | Activity | Details |
|---|---|---|
| 1 | Geometry creation | Generate solid parts of casting, risers, gates, chills in the pre-processor |
| 2 | Mesh generation | Apply finite difference cells with proper refinement to capture thin walls and hot spots |
| 3 | Property definition | Assign material thermal-physical properties, heat transfer coefficients, and boundary conditions |
| 4 | Filling simulation | Solve Navier-Stokes equations for mold filling, track free surfaces, and obtain initial temperature distribution |
| 5 | Solidification simulation | Solve nonlinear heat conduction with latent heat release, calculate solid fraction evolution |
| 6 | Defect prediction | Apply shrinkage criteria (e.g., Niyama, temperature gradient, solid fraction) to identify defective zones |
| 7 | Stress simulation | Perform thermal stress analysis to predict hot tears and residual stress |
| 8 | Result visualization | Display temperature fields, velocity fields, defect distribution, and stress contours |
3. Manufacturing Process Preparation for Box-Type Steel Castings
3.1 Electric Arc Furnace Steelmaking
The production of high-quality steel castings starts with the melting process. In my workshop, an electric arc furnace (EAF) is used to melt the charge, which consists of scrap steel, return scrap, and alloying additives. The EAF steelmaking can be divided into five stages: raw material collection, preparation for melting, melting period, oxidizing period, and reducing period.
3.1.1 Raw Materials
Scrap steel must be clean, free of rust, oil, and non-ferrous metals. The sulfur and phosphorus contents should be low to facilitate refining. The scrap size should be suitable for rapid melting without bridging. Additionally, some pig iron may be added to raise the carbon content, typically not more than 10% of the charge.
3.1.2 Melting and Refining
During the melting period, the electric arc melts the charge while a basic slag is formed to protect the steel from oxidation and to remove phosphorus. The oxidizing period aims to reduce carbon by injecting oxygen or adding iron ore, which generates a carbon monoxide boil that helps to remove dissolved gases and non-metallic inclusions. After sufficient oxidation, the slag is removed and a reducing period is carried out to deoxidize and desulfurize the steel. Deoxidation is achieved by adding carbon powder, ferrosilicon, and aluminum. The final deoxidation is often accomplished by plunging aluminum into the steel stream during tapping.
The temperature control is critical. For carbon steel castings, the tapping temperature is usually 1600–1650°C. After tapping, the steel is allowed to settle in the ladle for at least five minutes to allow inclusions to float out before pouring.
3.2 Molding Materials and Core Sand Technology
For box-type steel castings, the mold and cores must possess sufficient high-temperature strength, thermal stability, collapsibility, and low gas evolution. In our foundry, the conventional sodium silicate-bonded sands have been replaced by ester-cured sodium silicate sand systems, which offer better collapsibility and reclamation characteristics.
3.2.1 Sodium Silicate Sand Properties
Ester-cured sodium silicate sand, also called “ester hardener sodium silicate sand,” consists of silica sand, modified sodium silicate (with a high modulus and low viscosity), and a glycerol acetate ester hardener. Typical proportions are given in Table 3-1.
| Component | Content (parts by weight) |
|---|---|
| Silica sand (AFS 45-55) | 100 |
| Modified sodium silicate | 2.5–3.5 |
| Ester hardener | 0.3–0.5 |
The ester hardener promotes a chemical reaction with sodium silicate, forming a silica gel binder that imparts strength to the mold. The hardening time can be adjusted by selecting fast or slow ester hardeners and by varying the amount. In practice, the strip time is controlled between 10 and 45 minutes depending on the ambient temperature. A typical mixing sequence is:
Raw sand + ester hardener → mix for 1 minute → add sodium silicate → continue mixing for 2 minutes → discharge.
3.2.2 Molding Equipment
For efficient production, I supervised the development of a dedicated automated line for ester-cured sodium silicate sand molding. The line includes sand preparation, sand temperature control, continuous mixers, vibration tables, roll-over draw machines, core drying ovens, and roller conveyors. The use of this line greatly improved dimensional consistency and reduced labor intensity. The measured tolerance of water glass addition is within ±1%, and the uniformity of mixing is excellent.
3.3 Quality and Casting Defects Analysis of Box-Type Castings
Box-type castings may exhibit various casting defects, each with characteristic features and causes. A thorough understanding of these defects is a prerequisite for process improvement. The major types encountered in my production practice are listed in Table 3-2.
| Defect Type | Appearance | Main Causes | Preventive Measures |
|---|---|---|---|
| Gas porosity | Rounded or elongated cavities with clean walls; often located near surface | Excessive gas evolution from mold, inadequate venting, high gas content in liquid steel | Reduce mold moisture, improve venting, control pouring temperature, degassing of steel |
| Sand inclusions | Irregular holes containing sand particles | Mold erosion, loose sand in cavity, weak mold surface | Increase mold strength, avoid turbulent filling, clean cavity before closing |
| Slag inclusions | Irregular non-metallic deposits on upper surfaces or under cores | Inadequate slag removal, secondary oxidation, dirty ladle | Use teapot ladles, filters, proper slag practice |
| Shrinkage cavity | Large irregular void at hot spots; rough dendritic walls | Insufficient feeding, improper riser size or location | Optimize riser design, use chills, directional solidification |
| Shrinkage porosity | Fine interdendritic pores, often in axial zones or thick sections | Low thermal gradient, poor feeding through dendritic network | Increase gradient, apply pressure risers, reduce alloy freezing range |
| Hot tearing | Irregular, oxidized crack along grain boundaries | Restrained contraction, high thermal stress, weak hot strength | Improve mold collapsibility, reduce restraint, avoid stress raisers |
| Cold cracking | Straight or stepped crack with shiny fracture surface | Residual stresses above fracture strength during cooling | Slow cooling, stress relief, avoid stress concentration |
| Mold penetration / burn-on | Adherent layer of sand fused to casting surface | High pouring temperature, coarse sand, low refractory | Use finer sand, coating, lower temperature |
The formation of shrinkage defects is closely related to the solidification characteristics. For steel, the total volumetric contraction from liquidus to solidus is approximately 5–7%. The liquid feeding behavior can be described by the continuity equation and Darcy’s law for flow through the mushy zone. A common mathematical model for shrinkage porosity prediction is based on the concept of critical solid fraction. The criterion for macrosegregation and microporosity formation can be expressed using the local thermal gradient \(G\) and cooling rate \(R\):
$$ \frac{G}{\sqrt{R}} < C_{crit} $$
where \(C_{crit}\) is a critical constant, typically 1 for steel, but it may vary with casting geometry and alloy composition. A smaller value of \(G/\sqrt{R}\) indicates a higher probability of microporosity. Therefore, to avoid casting defects, the riser must maintain a high temperature and a steep thermal gradient towards the riser during solidification.
4. Case Study 1: Axle Box Housing with Cracks
4.1 Product Description and Initial Casting Process
The axle box housing is a semi-open cylindrical box structure, manufactured from low-alloy steel, with a gross weight of about 320 kg. It serves as a key component in a semi-suspended high-speed bogie. Due to the alternating loads at the root of the two end flanges, it is very susceptible to fatigue crack initiation. Thus, the technical specification demands zero casting defects in these regions.
The initial casting process design used a horizontally parted two-part mold with a middle gating system. Two open risers were placed at the top of the end flanges, while four insulating blind risers were placed at the bolt pad positions. The ingates were located at the intersection of the parting line and the bolt pads, with pouring from both ends. A three-dimensional schematic of the original layout is described as follows: the casting has a central cylindrical body, two end flanges, and bolt bosses. The gating enters from the side and fills the cavity upward.
4.2 Detection of Casting Defects
During the first production run, after machining, nearly half of the castings revealed fine cracks near the fillet radius between the parting line and the bolt bosses. Further destructive examination showed that these fine cracks were accompanied by large internal shrinkage cavities beneath the surface. The cracks and cavities were located in a thick section between the open riser and the blind riser. The presence of these casting defects seriously endangered the fatigue life of the component and resulted in a major quality accident.
4.3 Root Cause Analysis
Through engineering analysis, I identified three primary causes of the casting defects:
4.3.1 Inappropriate Ingate Location
The ingates were placed on the outer side of the end flange risers. This arrangement created a large hot spot at the junction because the ingate thickness added extra metal and heat. Moreover, the hot spot was outside the feeding range of the adjacent risers. The steel in the hot spot actually fed the cooler lower flange, causing a deep shrinkage cavity in the hot spot region.
4.3.2 Inadequate Riser Feeding Distance
The end open risers could not adequately feed the lower half of the flange, because the neck or riser connection was too small and the feeding path was blocked by the narrow cross-section at the parting line. The temperature field simulation showed a high temperature region in the lower flange that was not effectively connected to the riser. Thus, the riser solidified earlier than the casting section it was intended to feed.
4.3.3 Excessive Stress Concentration
During solidification, the two large end risers contracted, but the central sand core and mold provided a rigid restraint at the flange roots. This resulted in a high tensile stress at the fillet, triggering hot tears. The combination of a shrinkage cavity and a hot tear was observed as a crack.
4.4 Numerical Simulation and Process Improvement
To quantitatively understand the stress and solidification behavior, I applied a finite-element stress solver coupled with thermal analysis. The model included the casting, risers, and mold. The material properties were taken from the database for the given steel grade. The pouring temperature was set to 1550°C, and the simulation tracked the evolution of temperature and stress until the casting had cooled to ambient temperature.
The stress field results indicated a maximum principal tensile stress of approximately 280 MPa in the fillet region, which exceeded the hot strength of the steel at the end of solidification. This confirmed the hot tearing tendency. The thermal field also showed that the shrinkage cavity formed in the hot spot because the local temperature remained above the liquidus for a long time while the feeding channels were blocked.
Based on the simulation, I proposed three modifications:
- Relocation of ingates: Move the ingates from the flange side to the lower part of the bolt blind risers. This eliminates the artificial hot spot and provides a direct feeding path from the blind riser to the flange.
- Improvement of riser connection: Enlarge the riser neck (riser seat) and increase the height of the open risers to improve the feeding gradient and extend the feeding distance.
- Addition of stress-relief transition blocks: Add temporary projections at the fillet roots to shift the stress concentration away from the critical fillet. These blocks are removed during final cleaning.
The improved process was simulated again. The resulting stress distribution showed a reduction of maximum tensile stress to about 160 MPa, below the crack initiation threshold. The temperature field at 90% solidification showed a proper temperature gradient from the flange toward the riser, ensuring that the riser remained hotter than the casting section. The shrinkage porosity criterion indicated no defective zones in the critical region.
4.5 Verification and Results
After implementing the improved process, several trial castings were produced and subjected to sectional examination. The results are summarized in Table 4-1.
| Parameter | Original Process | Improved Process |
|---|---|---|
| Shrinkage cavity depth (mm) | 15–25 | 0 |
| Crack length (mm) | 10–20 | 0 |
| Feeding distance from riser (mm) | 80 | 140 |
| Maximum residual stress (MPa) | 280 | 165 |
| Yield of castings | 55% | 98% |
The improved casting process completely eliminated the shrinkage cavities and hot tears at the fillet roots. This solved the recurring quality problem and was adopted for mass production.
5. Case Study 2: Bracket for High-Power Locomotive
5.1 Product Characteristics and Casting Challenges
The bracket is a critical component of the bogie frame of a high-power diesel locomotive, co-produced with a U.S. company. The bracket weight is 85 kg, and the material is a low-alloy cast steel equivalent to ASTM A148 grade 80-50. The product has a generally thin plate-like structured geometry, except for a stepped boss of about 120 mm diameter where the thickness increases abruptly. This sharp change in section thickness created a high risk of internal shrinkage defects. Since the bracket undergoes significant load during service, any internal casting defect would be unacceptable.
5.2 Initial Process Design (Design A)
The initial casting design used a two-part molding with a horizontal parting line, a middle-pouring gating system, and ester-cured sodium silicate sand for both mold and cores. Two open risers were placed on the upper surface of the boss region. Owing to space constraints, the riser diameters were set to 100 mm and 120 mm, and the heights were limited by the available flask height to 150 mm for both. The gating system consisted of a sprue and two runners entering through the side walls. The initial design aimed to use the open risers to feed the boss. However, due to the narrow top surface, no chills or padding could be placed near the boss, making the directional solidification difficult to achieve.
5.3 Simulation of Process A
A three-dimensional model was built using the pre-processor, and the simulation was run under the following conditions:
- Pouring temperature: 1560°C
- Filling time: 8 seconds
- Mold material: sodium silicate sand
- Alloy: low-alloy cast steel
The temperature field at the end of solidification (95% solid) is shown conceptually here. The simulation revealed that the temperature in the open risers was lower than that in the boss region at the critical moment, indicating that the risers cooled faster than the boss. As a consequence, the risers lost their feeding ability and the boss solidified with a large internal shrinkage cavity. The solidification sequence was not directional; instead, isotherms were parallel to the surface and moved toward the center, causing simultaneous solidification across the thick section. When the mushy zone collapsed, feeding through the dendrites was blocked, producing axial shrinkage porosity.
The simulation predicted a distinct shrinkage cavity of about 12 mm in diameter within the boss, as shown in the visual results. This was later confirmed by casting trial sectioning, which indeed showed a severe shrinkage hole in the center of the boss.
5.4 Process Improvement: Insulating Sleeves (Design B)
Since the open riser dimensions could not be increased, I decided to replace the open risers with insulated (hot-topped) risers to reduce the cooling rate at the risers. The insulating sleeves were made of a ceramic fiber material with a much lower thermal conductivity than sand. The modified riser dimensions were reduced to 90 mm and 110 mm in diameter, with the same height of 150 mm, because the insulating sleeves improved the feeding efficiency. The process was simulated again under identical conditions.
The temperature field after the change showed that the riser portions retained a distinctly higher temperature than the boss, allowing a proper temperature gradient. The Niyama criterion distribution indicated that the axial shrinkage porosity was considerably reduced, but the simulation still predicted a narrow band of microporosity at the core of one of the bosses. Since this residual defect was very small and not likely to affect performance, I decided to further improve the riser design.
5.5 Final Optimization (Design C)
In the final design, the insulating riser dimensions were increased to 110 mm and 130 mm in diameter, while the height remained 150 mm. The added sleeve thickness gave a larger effective feeding volume. The simulation results for the final design showed that the risers stayed hot much longer than the casting, and the shrinkage porosity completely disappeared. The riser neck was also redesigned with a larger contact area to promote feeding. The predicted residual porosity was zero, and the stress distribution was acceptable.
5.6 Production Verification
To confirm the simulation predictions, a trial batch of brackets was produced using the final optimized riser design. The castings were then sectioned along the boss centerline for macroscopic examination. The sectioning revealed a completely sound structure with no visible shrinkage cavities or porosity. Based on these results, the final casting method was implemented for the production of the brackets. Table 5-1 compares the simulation predictions and the actual results for each design iteration.
| Design | Riser Type | Riser Size (mm) | Simulated Shrinkage | Actual Trial Result |
|---|---|---|---|---|
| A | Open | 100×150, 120×150 | Large cavity (12 mm) | Large shrinkage cavity |
| B | Insulated | 90×150, 110×150 | Small porosity band | Fine porosity (visible) |
| C | Insulated | 110×150, 130×150 | No shrinkage | Sound casting |
The close agreement between simulation predictions and actual casting defects validated the accuracy of the simulation methodology for predicting casting defects. The study demonstrated that numerical simulation is an invaluable tool for optimizing the feeding system and eliminating internal porosity in box-type steel castings.
6. Discussion on Formability of Box-Type Steel Castings
Box-type castings exhibit unique solidification characteristics due to their closed or semi-closed structures. The presence of internal cores often creates constraints that hinder contraction, increasing the risk of hot tearing and residual stress. The transition from thin walls to thick bosses or flanges also favors the formation of isolated hot spots. My experience with the axle box housing and the bracket illustrates that the following aspects are critical for ensuring soundness of box-type castings:
6.1 Gating and Ingate Placement
The gate location should not create local hot spots. Ideally, ingates should be positioned near risers or through the riser itself so that the incoming liquid steel preheats the riser and maintains a favorable temperature gradient. In my first case, moving the ingate away from the flange root had a dramatic effect on reducing shrinkage defects. A similar principle was applied to the bracket, where the gating was placed along the central parting line to avoid premature freezing of the riser neck.
6.2 Riser Design and Insulation
Riser sizing based on the modulus method is common, but for box-type castings with limited top surface, insulating sleeves offer a flexible solution. The use of exothermic or insulating materials increases the effective feeding distance and allows smaller riser footprints. The simulation proved that simply changing from open risers to insulated risers was not enough; the size had to be optimized further to fully suppress casting defects. The final success was achieved when the riser’s modulus exceeded the modulus of the casting junction times a safety factor. The modulus calculation is:
$$ M = \frac{V}{A} $$
where \(M\) is the modulus (solidification time characteristic), \(V\) is the volume of the metal, and \(A\) is the surface area in contact with the mold. For a riser to feed a casting, the riser modulus \(M_r\) must be larger than the casting modulus \(M_c\) by a factor of about 1.1 to 1.2. Insulating sleeves reduce the effective surface area \(A\) for heat loss, thereby increasing the apparent modulus.
6.3 Stress and Hot Tear Prevention
Hot tears are often caused by a combination of thermal gradient, mechanical restraint, and the high-temperature brittle range. The numerical stress simulation allowed me to visualize the stress concentration zones and to design stress-relief blocks added temporarily to the casting geometry. These blocks absorbed the deformation and shifted the maximum stress away from the critical fillet. After solidification, the blocks are cut off, leaving a sound structural fillet.
6.4 Use of Simulation Criteria
In this study, the following criteria were employed to predict casting defects:
- SHRINKAGE POROSITY: The Niyama criterion \(G/\sqrt{R}\) (also written as \(G/\sqrt{T_c}\)) where \(T_c\) is the local cooling rate. A threshold value of 1.0 is commonly used for steel; values below this indicate porosity.
- HOT SPOT: The temperature gradient criterion, \(G\), and the temperature gradient per unit solid fraction.
- FEEDING TIME: The time during which the metal is liquid above a critical fraction.
- HOT TEAR INDEX: A function that combines thermal strain and strain rate over the brittle temperature range.
The validity of these criteria depends on the alloy system and the solidification mode. For low-alloy steel, which solidifies with a dendritic morphology, the Niyama criterion gave accurate predictions when the simulation mesh was fine enough. In the axle box housing, a coupling between the thermal and stress solver was necessary to predict hot tears. The stress solver used an elasto-viscoplastic constitutive model with temperature-dependent mechanical properties.
7. Conclusions
Through the investigation of two typical box-type steel castings, I drew the following conclusions:
- The formability of box-type steel castings is largely governed by the proper design of gating and riser systems. Inappropriate ingate placement and inadequate feeding paths are the main causes of shrinkage-related casting defects. Numerical simulation can accurately predict the location and size of these defects before production.
- For the axle box housing, the casting defects were eliminated by relocating the ingates to the blind risers, enlarging the riser seats, and adding stress-relief transition blocks. The improved process resulted in zero internal cavities and zero hot tears, verified by destructive inspection.
- For the bracket, the initial open riser design could not provide a favorable thermal gradient. Replacing them with insulated risers and increasing the riser size to 110×150 mm and 130×150 mm completely eliminated shrinkage porosity in the thick boss. The simulation predictions were in good agreement with actual trial results.
- The use of finite-element stress analysis in combination with thermal-solidification simulation is essential for hot tearing prediction in restrained box-type castings. The maximum principal stress criterion successfully identified the crack-prone zones and guided the modification of the casting geometry to reduce stress concentration.
- Casting defect prediction via numerical simulation has been proven to be a reliable and cost-effective method for process optimization. It significantly reduces the need for costly trial runs and shortens the product development cycle. The insights gained from this study are applicable to other similar railway components.
In summary, this study demonstrates that systematic analysis of casting defects, supported by numerical simulation tools, can yield robust manufacturing processes for box-type steel castings. The combination of process knowledge and simulation provides a scientific basis for defect elimination, leading to higher quality, lower cost, and safer railway products.
In the future, I plan to extend this methodology to more complex multi-core box castings and to incorporate microstructural modeling for quantitative prediction of mechanical properties. The continuing development of simulation software will further improve the accuracy of casting defect prediction and contribute to the digital transformation of foundry practices.