In this thesis, I systematically investigate the formability and casting defects of box-type steel castings, focusing on two critical components: the axle box housing and the bracket used in railway locomotives. The study integrates foundry practice with advanced numerical simulation tools, specifically using ProCAST and MAGMAsoft software. By analyzing the filling, solidification, and stress fields during the casting process, I developed optimized casting process designs that effectively eliminate or minimize casting defects such as shrinkage porosity, hot tearing, and cold cracking. The research demonstrates that computer-aided simulation is an indispensable tool for modern casting process design, enabling engineers to predict and prevent casting defects before physical trials, thereby reducing costs and lead times while improving product quality and reliability.
1. Introduction
The casting industry serves as the foundational sector supplying semi-finished castings to various manufacturing industries, and it plays a vital role in the national economy. With the rapid development of railway transportation, particularly the “heavy haul freight and high-speed passenger” initiatives, the requirements for locomotive components have become increasingly stringent. Box-type castings, such as main bearing housings and axle box housings, are widely used in locomotive diesel engines and bogie systems. The quality of these castings directly influences the overall performance and safety of locomotives. Casting defects, including shrinkage cavities, gas porosity, sand inclusions, and hot tears, are major concerns that can lead to catastrophic failures in service.
Traditional trial-and-error methods for casting process development are time-consuming and costly. With the advent of computer simulation technology, it is now possible to virtually predict the evolution of temperature fields, flow fields, and stress fields during mold filling and solidification. By applying appropriate criteria, casting defects can be forecasted, and process parameters can be optimized in the virtual environment. This study focuses on two representative box-type steel castings: the axle box housing (a semi-suspended high-speed locomotive component) and the bracket (a key part of a high-power diesel locomotive bogie). Both components are subject to high dynamic loads and require defect-free internal quality. I utilized ProCAST (with its stress module) and MAGMAsoft (with its thermal and feeding modules) to simulate and analyze the casting processes, leading to significant improvements in process design and defect elimination.
2. Overview of Casting Numerical Simulation
2.1 Significance of Numerical Simulation
Numerical simulation of casting processes has evolved from academic research to industrial application over the past several decades. The primary objective is to solve the problem of casting process optimization. The relationship between numerical simulation and process design can be summarized as follows:
- Solidification feeding simulation ↔ riser, chill, and padding design
- Mold filling simulation ↔ gating system design
- Solidification structure simulation ↔ microstructure and mechanical property optimization
- Thermal stress/strain simulation ↔ hot tearing prevention
By simulating the complex physical phenomena during casting, engineers can obtain three-dimensional images of temperature, velocity, stress, and concentration fields. Combined with defect prediction criteria, these simulations enable:
- Prediction of solidification time and productivity
- Prediction of shrinkage porosity and macro-shrinkage
- Prediction of mold surface temperatures for die design
- Control of solidification conditions
- Provision of data for stress, segregation, and property analyses
2.2 Historical Development
Research on solidification simulation began in the 1960s. The first finite difference method applied to solidification was reported by G. F. Weiner and others. In the 1970s, many industrial countries initiated intensive research. By the 1980s, commercial software emerged, and in the 1990s, with the rapid growth of computing power, casting simulation became a practical tool. In China, research started in 1978, and now domestic software such as FT-Star, Huazhu CAE, and others are available. International major products include ProCAST, MAGMASOFT, Flow-3D, and SolidCast.
2.3 Typical Simulation Software and Workflow
Among the various software packages, ProCAST is a well-known finite element solver that can simulate heat transfer, fluid flow, stress, and microstructure. Its workflow consists of pre-processing, meshing, simulation setup, and post-processing. I used ProCAST for stress analysis of the axle box housing, and MAGMASOFT for thermal and shrinkage analysis of the bracket. The general steps are:
- Create solid models in Unigraphics (UG) and export as IGS or similar files, separating casting, risers, gates, and chills.
- Import into simulation software, assign material properties and boundary conditions.
- Generate mesh (finite element for ProCAST, finite difference for MAGMAsoft).
- Set process parameters: pouring temperature, pouring time, heat transfer coefficients, etc.
- Run simulation and visualize results: temperature, velocity, solid fraction, stress, and defect criteria.
A key criterion used in shrinkage prediction is the Niyama criterion (G/√Ṫ), where G is the thermal gradient and Ṫ is the cooling rate. The Niyama parameter is defined as:
$$\mathrm{Niyama} = \frac{G}{\sqrt{T}}$$
where \(G = |\nabla T|\) (K/mm) and \(T = \partial T/\partial t\) (K/s). When the Niyama value is below a critical threshold, shrinkage porosity is expected. For steel castings, the critical value often ranges from 0.5 to 1.0 depending on section size.
Another criterion is the temperature gradient at the solidification front, and the feeding resistance criterion. In my research, I used the porosity (Poroclose) and H-Mat (hot spot) criteria available in MAGMASOFT.
3. Foundry Process Preparation for Box-Type Steel Castings
3.1 Electric Arc Furnace Steelmaking
The production of high-quality steel castings requires careful control of the steelmaking process. In my foundry, an electric arc furnace (EAF) is used. The process comprises five stages: raw material preparation, melting, oxidation, reduction, and tapping.
| Stage | Main Tasks | Key Parameters |
|---|---|---|
| Raw materials | Selection of clean scrap, no non-ferrous contamination | S & P < 0.05% |
| Melting | Fast melting, early slagging for dephosphorization | Power input, O₂ lancing |
| Oxidation | Decarburization, degassing, heating | Temperature > 1500°C, decarbonization rate > 0.01%/min |
| Reduction | Deoxidation, desulfurization, alloying | White slag, C₂Ca formation, final Al addition |
| Tapping | Clean steel, temperature control | Steel temperature 30–50°C above liquidus |
During reduction, the slag is made white by adding carbon powder and calcium carbide. This generates a reducing atmosphere, enabling efficient deoxidation and desulfurization. Finally, aluminum is added for final deoxidation during tapping, and the steel is teemed after a settling time.
3.2 Molding Materials and Process
For steel castings, the mold material must withstand high temperatures and provide adequate collapsibility. Traditionally, sodium silicate (water glass) sand was used. I improved the process by adopting ester-cured sodium silicate sand, which offers better collapsibility and reclamation properties compared to conventional CO₂-silicate sand.
| Property | Conventional CO₂-silicate sand | Ester-cured silicate sand |
|---|---|---|
| Binder content | 6-8% | 2.5-3.5% |
| Hardening time | Fast (CO₂) | Adjustable (10-60 min) |
| Collapsibility | Poor | Good |
| Reclamation | Difficult | Feasible |
| Surface finish | Reasonable | Excellent |
The sand mixture consisted of silica sand (AFS 45-55), modified sodium silicate (modulus 2.4-2.6, density 1.48), and glycerol acetate ester as hardener. The typical proportions are:
$$\text{Sand (100 parts)} + \text{Silicate (2.8-3.5%)} + \text{Ester (0.2-0.4%)}$$
The mixing was performed using a continuous twin-arm mixer. The molding line included a jolt-squeeze machine and a turnover-drawing machine. This process produced molds with high dimensional accuracy and good surface quality, reducing the incidence of sand-related casting defects.
3.3 Common Casting Defects in Box-Type Castings
I analyzed the most frequent casting defects encountered in steel castings, including gas porosity, sand inclusions, slag inclusions, shrinkage cavities, hot tears, and cold cracks. Their causes and prevention methods are summarized below.
| Defect type | Characteristics | Formation mechanism | Typical countermeasures |
|---|---|---|---|
| Gas porosity | Round/elongated holes, smooth walls | Gas evolution from mold or dissolved gases | Reduce gas evolution, improve venting, higher pouring temp |
| Sand inclusions | Irregular cavities containing sand | Mold erosion or loose sand | Stronger mold surface, lower erosion velocity |
| Slag inclusions | Non-metallic particles, often near top surfaces | Slag from melting or oxidation | Clean steel, slag filters, bottom pouring |
| Shrinkage cavity | Large irregular hole at hot spots | Inadequate feeding, localized hot spot | Proper risers, chills, directional solidification |
| Shrinkage porosity | Dispersed small voids | Feeding channels blocked, insufficient gradient | Reduce hot spots, enhance gradients, use chills |
| Hot tearing | Irregular, oxidized fracture surfaces | Restrained contraction during solidification | Improve mold collapsibility, avoid sharp corners |
| Cold cracking | Straight, clean fracture, often after cooling | Excessive residual stresses | Slow cooling, stress relief annealing |
In particular, for box-type steel castings with complex geometries and varying wall thicknesses, the combination of shrinkage porosity and hot tearing is the most challenging. Numerical simulation proved to be the most effective approach to predict these casting defects and optimize the process.
4. Case Study I: Axle Box Housing – Crack Defect Improvement
4.1 Product and Initial Casting Process
The axle box housing is a key component of the semi-suspended high-speed locomotive bogie. It was made of ZG25MnNiV steel (similar to ASTM A148 grade 105-85), weighing approximately 400 kg. The component has a semi-open cylindrical structure with two flanges at the ends. Owing to the alternating loads at the flange roots, the technical requirements strictly prohibit any internal casting defects. The initial casting process used a horizontal split (two-box) with a mid-plane gating system. Two open risers were placed on the flanges, and four blind risers were placed on the bolt bosses. The ingates were positioned at the junction of the flange and bolt boss on the parting line. Figure below shows the initial three-dimensional process model.
During the first batch production, nearly 50% of the castings exhibited small hairline cracks at the root of the circular flange near the bolt boss after machining. Further ultrasonic inspection revealed large shrinkage cavities in the interior of the same region.
4.2 Defect Analysis
The defects occurred between the open and blind risers. The presence of the ingate increased the local thermal mass, creating a large hot spot. Moreover, the hot spot was located at the dead zone outside the feeding range of both risers. As solidification proceeded, the open riser contracted, but its contraction was hindered by the sand mold between the flanges, generating high tensile stress at the flange root. This stress, combined with the weakened section due to shrinkage cavity, led to crack formation. The crack was a classic hot tear during solidification, aggravated by the large shrinkage cavity. Because all castings shared the same gating design, even those not showing cracks after machining likely contained internal voids, posing a serious safety hazard.
4.3 Process Improvement
I used ProCAST software to model the casting process and perform a finite element stress analysis. The simulation included the temperature field, solidification sequence, and the evolution of thermal stress and residual stress. From the simulation results, I identified three major issues:
- Ingate location was unsuitable, creating an additional hot spot.
- The risers could not feed the lower part of the flange because of the narrow mold section at the parting line.
- Stress concentration was high at the flange root due to restrained contraction.
Based on these findings, I implemented the following modifications:
- Moved the ingate from the flange root to the lower part of the blind risers, eliminating the original hot spot and improving feeding.
- Added riser pedestals (pads) under the open risers to increase their feeding range and create a smoother thermal gradient.
- Added temporary stress-relief transition blocks at the flange roots, which effectively moved the stress concentration point away from the critical section. The transition blocks were removed during fettling after solidification.
The stress simulation after modification showed a significant reduction in residual stress at the flange root. Figure shows the residual stress distribution predicted by ProCAST. The color map indicates that the maximum stress has shifted away from the critical corner to the transition block, which is later removed.
After the process modification, trial castings were sectioned. The micro-etching and dye penetrant test confirmed that no shrinkage cavities were present in the previously defective zone. The cracks disappeared completely. This case validates that ProCAST stress analysis is an effective tool for preventing hot tearing in box-type steel castings.
4.4 Summary of Axle Box Housing Improvement
| Parameter | Original | Modified |
|---|---|---|
| Ingate position | Flange root (outside riser) | At lower part of blind riser |
| Open risers | Directly on flange | With riser pad |
| Transition block | None | Added at flange root |
| Shrinkage cavity | Large internal | None |
| Crack | Present in 50% of castings | Eliminated |
5. Case Study II: Bracket Casting – Simulation and Process Optimization
5.1 Product and Initial Process
The bracket is a key component of the high-power diesel locomotive bogie produced in cooperation with a US company. Its weight is 450 kg, and the material is CAST ZG25MnNiV (equivalent to ASTM A148 105-85). The structure is relatively complex, with a large flat plate and a suddenly increased thickness at a circular boss of diameter φ340 mm. Because of this abrupt section change, the risk of shrinkage porosity at the boss is high. The initial casting design employed a two-box split pattern, a mid-plane gating system, and ester-cured sodium silicate sand. Two open risers were placed over the bosses, with dimensions of φ240 mm × 300 mm and φ220 mm × 300 mm. The riser sizes were calculated using the proportional method, but the mold height limited the riser height.
5.2 Simulation of Initial Design (Scheme A)
I used MAGMASOFT to simulate the filling and solidification. The process parameters were: pouring temperature 1550°C, pouring time 15 seconds. The temperature field at 80% solidification is shown in Figure (not cited). It became clear that the riser temperature was lower than the temperature of the boss underneath. This indicates that the risers cooled faster than the casting hot spot, violating the requirement for directional solidification. The feeding effect was thus insufficient, and the Niyama criterion map revealed a local shrinkage porosity along the axis of both bosses.
The Niyama criterion values in the defective region were lower than 0.5, confirming shrinkage porosity. The reason was the open riser has a large surface area and no thermal insulation, so it lost heat rapidly. Even though the riser volume was large, the feeding efficiency was low. The results proved that the open riser design could not produce sound castings.
5.3 Modification 1: Replace Open Risers with Insulated (Sleeved) Risers
Because the structural constraints prevented increasing the riser dimensions, I replaced the open risers with insulated risers (using exothermic/insulating sleeves). The new riser sizes were φ200 mm × 250 mm and φ180 mm × 250 mm, which were smaller in volume but had much lower cooling rates due to the insulation. The simulation of Scheme B showed that at 80% solidification, the riser temperature was slightly higher than that of the boss, indicating an improved thermal gradient. The Niyama criterion showed that the severe shrinkage porosity disappeared, but a small residual shrinkage remained in the right-hand boss center. In addition, there was a risk of shrinkage cavity at the riser neck because the insulation enhanced the hot spot at the flange connection.
The remaining small shrinkage porosity was considered acceptable by design staff, but I aimed for further improvement. The absolute defect size was reduced, but not fully eliminated.
5.4 Modification 2: Further Increase Riser Sizes with Insulation
To fully eliminate the residual shrinkage porosity, I modified the insulated riser dimensions to φ230 mm × 280 mm and φ210 mm × 280 mm. The simulation Scheme C showed that at 80% solidification, the risers are significantly hotter than the boss (temperature difference > 50°C). The Niyama criterion indicated that no porosity exists in the boss interior. The shrinkage cavity in the riser is located far from the casting surface, ensuring soundness.
| Scheme | Riser type | Riser size (mm) | Simulation result | Actual casting result |
|---|---|---|---|---|
| A | Open | φ240×300, φ220×300 | Large shrinkage porosity in boss | Rejected: serious shrinkage |
| B | Insulated | φ200×250, φ180×250 | Small residual shrinkage porosity | Accepted? but minor porosity |
| C | Insulated | φ230×280, φ210×280 | No shrinkage porosity | Accepted, defect-free |
5.5 Verification by Production Trial
After I adopted Scheme C, the foundry produced two trial castings. Both were sectioned at the boss area and subjected to ultrasonic and dye penetrant inspection. No shrinkage cavities or any other casting defects were found. The actual results matched the simulation predictions very well. This confirms the reliability of MAGMASOFT in predicting casting defects for box-type components.
Additionally, the simulations allowed me to calculate the complete feeding process. The feeding efficiency of the insulated riser is substantially higher than the open riser. The heat loss from the open riser’s top surface can be expressed as:
$$Q = hA(T_r – T_{amb}) + \varepsilon \sigma A (T_r^4 – T_{amb}^4)$$
where \(h\) is the convective heat transfer coefficient, \(A\) is the surface area, \(\varepsilon\) is emissivity, and \(\sigma\) is the Stefan-Boltzmann constant. By covering the riser with an insulating sleeve, the top radiation and convection losses are minimized, and the sidewall heat transfer is reduced, resulting in a lower cooling rate. The temperature distribution during solidification obeys the Fourier equation:
$$\rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \dot{q}$$
where \(\rho\) is density, \(c_p\) specific heat, \(k\) thermal conductivity, and \(\dot{q}\) the latent heat source term. The simulation solves this equation with appropriate boundary conditions.
6. General Rules for Box-Type Steel Casting Process Design
Through the two case studies, I have established a set of practical guidelines to prevent casting defects in box-type steel castings:
- Avoid ingate positions that create isolated hot spots. Ingates should be placed near risers to allow feeding to be effective.
- Ensure a proper thermal gradient. The riser must solidify last. Use insulation or exothermic sleeves to delay riser freezing.
- Design risers based on modulus and feeding distance. The modulus of the riser should be at least 1.2 times the modulus of the hot spot. The simulation can help fine-tune the riser dimensions.
- Use chills to accelerate cooling of thick sections that are difficult to feed. In the axle box housing, chills on the lower flange could reduce the hot spot, but the stress consideration led to the addition of transition blocks.
- Consider thermal stress and hot tearing. Box-type castings with long flanges are prone to hot tearing at roots. Add stress-relief features or make the mold more collapsible.
- Validate the process with simulation before production. The use of Niyama criterion and stress maps provided accurate predictions consistent with physical trials.
These rules are applicable to similar box-type components such as gearboxes, bearing housings, and structural brackets. The combination of ProCAST and MAGMASOFT offers a complete solution for thermal, feeding, and stress analysis.
7. Conclusion
In this thesis, I systematically studied the formability and casting defects of box-type steel castings. The following conclusions can be drawn:
- By analyzing the gating and risering systems of the axle box housing and the bracket, I identified critical design flaws that caused severe shrinkage cavities and hot tears. The root causes were improper ingate positions, inadequate feeding ranges, and excessive stress concentrations.
- Using ProCAST stress analysis, I successfully simulated the residual stress distribution and predicted the hot tearing tendency. The addition of a transition block at the flange root proved to be an effective method to reduce stress concentration and eliminate cracking.
- Using MAGMASOFT thermal and feeding analysis, I optimized the riser design for the bracket. The simulation showed that replacing open risers with insulated risers of appropriate size achieved directional solidification and eliminated shrinkage porosity at the critical boss. The trial production confirmed the simulation accuracy.
- The Niyama criterion, defined as \(G/\sqrt{\dot{T}}\), is a reliable indicator for predicting shrinkage porosity in steel castings. For the bracket, the critical Niyama value was about 0.5; values below this threshold produced porosity.
- The numerical simulation approach drastically reduced the product development costs and lead time. It also improved the quality and reliability of castings, ensuring safe operation in railway applications. The methodologies and guidelines developed in this work are beneficial for future box-type steel casting production and contribute to the advancement of intelligent casting process design.
The results demonstrate that the integration of casting simulation with process engineering is a powerful tool to combat casting defects. I will continue to apply these techniques to other new products and further refine the simulation models to include microstructure predictions and multi-physics coupling. The ultimate goal is to achieve “zero-defect” casting production and to promote the digital transformation of the foundry industry.