The production of large steel castings has always been a challenging task due to the complex solidification behaviour, the occurrence of shrinkage porosity, hot tearing, gas entrapment and sand inclusion. These defects significantly undermine the mechanical integrity and service life of the final components. In the framework of a national key scientific and technological project, this study focuses on the numerical simulation of the casting process of a large steel castings component, namely the front beam of a 20,000-ton horizontal extruder. The main objective is to establish a reliable and repeatable simulation‑driven procedure for the design of casting processes of large steel castings. Using the finite element software ProCAST, a comprehensive analysis of the temperature field, flow field and stress field during the filling and solidification stages was carried out. The formation mechanisms of typical defects in large steel castings were investigated, and an improved casting process was developed to eliminate these defects. This paper presents the theoretical background, simulation methodology, key results and process optimisation strategies for large steel castings based on numerical simulation.
1. Introduction and Research Significance
Large steel castings are fundamental components for heavy machinery, such as extruder front beams, rolling mill housings and turbine casings. Their manufacturing quality directly determines the operational reliability and safety of the entire equipment. However, the casting process of these large steel castings is extremely difficult to control. The large dimensions and thick wall sections often lead to uneven cooling, significant thermal gradients and complex feeding behaviour, which consequently result in macroscopic defects such as shrinkage cavities, micro‑porosity, hot cracks, inclusions and distortion. Traditionally, foundry engineers relied on empirical rules and time‑consuming trial‑and‑error procedures to develop casting processes for large steel castings. This approach is both costly and inefficient, particularly for one‑of‑a‑kind heavy components.
In recent years, computational simulation has become a powerful tool to understand and optimise casting processes. Through the numerical modelling of the filling, solidification and cooling stages, engineers can visualise the evolution of temperature, flow and stress fields and predict the formation of defects in steel castings before any physical trial. This dramatically reduces the product development cost, shortens the lead time and improves the quality of steel castings. In this study, the front beam of a large horizontal extruder was selected as the research object. The component weighs approximately 365 tonnes and has a wall thickness ranging from 250 mm to more than 1,500 mm, which makes it an ideal representative of typical large steel castings. The paper systematically presents the finite element simulation of the casting process for this large steel castings component, in terms of the solidification temperature field, the mould filling flow field and the stress field during solidification, together with the corresponding optimisation of the casting process.
2. Theoretical Basis for the Temperature Field Analysis of Steel Castings
The solidification of steel castings involves complex heat transfer phenomena, including conduction, convection and radiation. In the simulation of the solidification of large steel castings, it is often assumed that the mould is filled instantaneously since the pouring time is very short compared with the total solidification time. Under this assumption, the temperature‑field calculation becomes the governing part of the analysis. The basic law for heat conduction is Fourier’s law, which is expressed as
$$q = – \lambda \,\nabla T$$
where \(q\) is the heat flux (W·m-2), \(\lambda\) is the thermal conductivity (W·m-1·K-1), and \(T\) is the temperature (K). The convective heat transfer between the surface of the casting and the mould is described by Newton’s law of cooling:
$$q = \alpha \left( T_f – T_w \right)$$
where \(\alpha\) is the heat transfer coefficient (W·m-2·K-1), \(T_f\) is the fluid temperature, and \(T_w\) is the solid wall temperature. At high temperatures, radiation also plays an important role, and the Stefan–Boltzmann law is adopted:
$$q = \varepsilon \sigma_0 T_s^4$$
with \(\varepsilon\) being the emissivity, \(\sigma_0 = 5.76 \times 10^{-8} \ \mathrm{W\cdot m^{-2}\cdot K^{-4}}\) the Stefan–Boltzmann constant, and \(T_s\) the absolute surface temperature.
For the finite element simulation, appropriate boundary conditions must be established. In this work, the interface between the steel castings and the sand mould is treated as a contact interface with a heat transfer coefficient of 500 W·m-2·K-1. The interface between the mould and the ambient air has a heat transfer coefficient of 10 W·m-2·K-1. Chills, when used, have a higher interface heat transfer coefficient of 1000 W·m-2·K-1 against the casting. Table 1 summarises the initial thermal parameters used in the simulation.
| Parameter | Casting (Chill) | Sand Mould | Air |
|---|---|---|---|
| Density (kg/m³) | 7800 | 1550 | 1.2 |
| Specific heat (kJ·kg⁻¹·°C⁻¹) | 0.829 | 1.09 | 1.00 |
| Thermal conductivity (W/(m·°C)) | 0.222 | 0.008 | 0.054 |
| Initial temperature (°C) | 1550 (pouring) | 25 | 25 |
3. Prediction of Shrinkage Porosity and Cavities in Steel Castings
Shrinkage cavities and porosity are among the most serious defects in steel castings. They arise from the volumetric contraction of the liquid and solidifying alloy that is not compensated by the remaining liquid metal. The solidification sequence, the feeding efficiency of risers and the local thermal gradient are the primary factors controlling the formation of these defects. In this study, three different criteria were used to predict shrinkage defects in large steel castings.
3.1 Critical Solid Fraction Method
This method assumes that the liquid metal can no longer flow when the solid fraction reaches a critical value. For steel castings, this critical value is typically taken as \(g_{sc}=0.7\). When isolated liquid regions are surrounded by a solid network with a solid fraction above this critical value, feeding is interrupted, and a shrinkage cavity or micro‑porosity will form. By plotting the solid fraction distribution during solidification, one can directly identify the locations of isolated liquid pools and hence the regions with a high risk of shrinkage defects in steel castings.
3.2 Temperature Gradient Criterion
The temperature gradient criterion is based on the observation that, at the end of solidification, insufficient thermal gradients indicate poor feeding conditions. For each element, the maximum temperature gradient with respect to its neighbours is evaluated:
$$G = \frac{\max \left( T_j – T_s \right)}{\Delta l}$$
where \(T_s\) is the solidus temperature and \(\Delta l\) is the distance between element nodes. If the value of \(G\) is below a certain threshold, a shrinkage defect is assumed to form. This criterion indirectly accounts for both thermal conditions and flow resistance in the mushy zone.
3.3 Niyama Criterion
The Niyama criterion is one of the most commonly applied prediction tools for shrinkage porosity in steel castings. It is based on the ratio between the local temperature gradient \(G\) and the cooling rate \(R\):
$$\frac{G}{\sqrt{R}} < C_{crit}$$
where \(C_{crit}\) is normally taken between 0.8 and 1.1 for steel castings, with the higher value being preferred for heavy‑section components. In the present study, a value of 1.1 was selected considering the massive dimension of the front beam. This criterion has been widely validated for both experimental and production steel castings and is implemented in the ProCAST software. The three prediction methods described above were simultaneously applied in the present simulation, and they yielded consistent results, which validated the robustness of the porosity prediction for the large steel castings component.
4. Material Properties and Modelling of ZG230-450 Steel Castings
The material selected for the front beam is ZG230‑450, a Chinese designation for a cast carbon steel with good strength and plasticity. The chemical composition is listed in Table 2.
| C | Si | Mn | P | S |
|---|---|---|---|---|
| ≤0.20 | ≤0.50 | ≤1.20 | ≤0.04 | ≤0.04 |
The thermodynamic properties of the material were calculated using the thermodynamic database of ProCAST. The liquidus and solidus temperatures were determined to be 1498 °C and 1435 °C, respectively. Figure 1 illustrates the temperature‑dependent properties of the ZG230‑450 alloy, including thermal conductivity, solid fraction, density and enthalpy. These data were introduced into the finite element model for the transient heat transfer calculations.
$$L=1498\,^\circ\mathrm{C},\quad T_s=1435\,^\circ\mathrm{C}$$
| Yield strength (MPa) | Tensile strength (MPa) | Elongation (%) | Reduction of area (%) | Impact energy (J) |
|---|---|---|---|---|
| 230 | 450 | 22 | 35 | 25 |
5. Initial Casting Process Design and Temperature‑Field Simulation
5.1 Geometry and Hot Spot Analysis
The front beam geometry was modelled using the UG three‑dimensional CAD software. Because of the large size of the component, only a quarter of the model was used for the finite element simulation to reduce computation time. In a first step, the casting without any risers was simulated to determine the locations of hot spots and isolated liquid regions. The simulation showed that the thick central section and the vertical plates of the front beam contained isolated liquid pools at the end of solidification, which would lead to shrinkage porosity in steel castings.
5.2 Riser Design for the Steel Castings
The casting was divided into four balanced zones according to the geometry and the distribution of hot spots. The modulus of each zone was calculated using the volume‑to‑surface‑area ratio:
$$M = \frac{V}{A}$$
The moduli of the different zones are given in Table 4. For the first and second zones, two large cylindrical open risers with a diameter of 2400 mm and a height of 2400 mm were designed. The feeding capability of each riser was verified using the formula:
$$G_{\text{cast}} = G_{\text{riser}} \frac{\eta – S}{S}$$
where \(\eta\) is the feeding efficiency of the riser (14% for ordinary sand risers) and \(S\) is the solidification contraction of steel (5.4%). For the vertical plate zones, three small insulated blind risers were used.
| Zone | Surface area (mm²) | Volume (mm³) | Mass (kg) | Modulus (cm) |
|---|---|---|---|---|
| Overall | 199,718,136 | 46,620,546,771 | 365,068 | 23.3 |
| 1 (thick) | 17,913,930,605 | 52,347,040,140 | 140,277 | 34.2 |
| 2 (thick) | 17,913,930,605 | 52,347,040,140 | 140,277 | 34.2 |
| 3 (plate) | 5,396,326,445 | 50,092,028 | 42,257 | 10.8 |
| 4 (plate) | 5,396,326,445 | 50,092,028 | 42,257 | 10.8 |
5.3 Simulation Results of the Initial Process
The solidification simulation of the initial process showed that the vertical plates contained isolated liquid regions during the later stage of solidification. The blind risers were not able to feed these regions adequately because the feeding channels were blocked by the solidified dendrite network. The thick sections, however, showed a proper feeding channel towards the open risers, suggesting that the large risers were capable of eliminating shrinkage defects in those regions. The overall prediction indicated the presence of shrinkage porosity in the plate areas, which was unacceptable for large steel castings of this importance.
6. Optimisation of the Casting Process and Improved Solidification
Based on the initial simulation results, the following corrective measures were proposed for the casting of the steel castings:
- The ordinary sand blind risers were replaced with insulated blind risers using a 70 mm thick insulation sleeve. The riser dimensions were accordingly reduced to B=700 mm, L=H=1050 mm.
- The two large cylindrical risers were changed to exothermic‑insulated risers with a 110 mm thick sleeve. The diameter was reduced to 1835 mm and the height to 2202 mm. A layer of exothermic covering agent was applied on the top of the molten steel in the open risers.
- To improve feeding of the vertical plates, wedge‑shaped padding was added at the outer side of the plates with a length of 300 mm.
- Eight indirect external chills were placed at the bottom of each vertical plate. Each chill had the dimensions 500 mm × 300 mm × 300 mm, and sand was pasted over the chill with a thickness of 10–15 mm to avoid an excessively strong chilling effect.
After the optimised process was implemented in the model, the revised solidification simulation was performed. The results showed a significantly improved temperature‑field distribution. The chills at the bottom of the vertical plates created an end‑effect that promoted directional solidification from the bottom towards the risers. The feeding channels remained open and no isolated liquid regions were found in the casting body. The shrinkage porosity predicted by the Niyama criterion was confined to the interior of the risers, which is the desired configuration for sound steel castings.
The cooling curves of selected points along the cross‑section of the large riser are shown in Figure 2. Point 4, located close to the chill, exhibited the fastest cooling rate, which confirmed the effectiveness of the chill in establishing the desired temperature gradient. Points 9 and 10, located inside the riser, cooled more slowly than the adjacent casting zones, ensuring that the riser remained liquid until the feeding was completed. The comparison between the initial and optimised processes is summarised in Table 5.
| Process | Total liquid steel mass (t) | Yield rate (%) | Casting quality |
|---|---|---|---|
| Initial | 610.5 | 59.79 | Defects in vertical plates |
| Optimised | 523.9 | 69.67 | No defects in the casting body |
The optimisation allowed the riser size to be significantly reduced while completely eliminating shrinkage defects in the steel castings. The yield rate increased from 59.79% to 69.67%, which represents a substantial material and cost saving for the production of large steel castings.
7. Numerical Analysis of the Mould Filling Process for Large Steel Castings
The filling process is the first stage of the formation of steel castings. An improperly designed gating system can lead to turbulent flow, air entrainment, mould erosion, cold shut and inclusions. For large steel castings, especially those with tall vertical dimensions, it is essential to design the gating system to ensure smooth, sequential and rapid filling. The flow of liquid metal in the gating system and the mould cavity can be described by the continuity equation and the Navier–Stokes equations:
$$\frac{\partial \rho}{\partial t} + \frac{\partial(\rho u_x)}{\partial x} + \frac{\partial(\rho u_y)}{\partial y} + \frac{\partial(\rho u_z)}{\partial z} = 0$$
$$\frac{\partial \mathbf{v}}{\partial t} + (\mathbf{v}\cdot\nabla)\mathbf{v} = -\frac{1}{\rho}\nabla p + \gamma \nabla^2 \mathbf{v} + \mathbf{g}$$
In addition to the flow equations, the design of the gating system for steel castings relies on hydraulic principles. The relationship between the filling time, the flow rate and the gating system cross‑section was derived from Bernoulli’s equation:
$$F = \frac{G}{\mu \gamma \tau \sqrt{2g H_{\text{avg}}}}$$
where \(F\) is the area of the minimum cross‑section in the gating system, \(G\) is the total weight of the liquid metal, \(\mu\) is the flow coefficient, \(\gamma\) is the specific weight, \(\tau\) is the filling time, and \(H_{\text{avg}}\) is the average hydraulic head. In the present work, the pouring was carried out using four ladles simultaneously, each with a 100 mm diameter nozzle and a flow rate of 190 kg/s. The total filling time was calculated as 690 s. The average rising speed of the liquid metal in the mould was:
$$v = \frac{H}{\tau} = \frac{5000}{690} \approx 7.3 \ \mathrm{mm/s}$$
This value satisfies the minimum rising speed requirement of 4 mm/s for simple‑shaped large steel castings.
8. Design of the Stepped Gating System for Steel Castings
A stepped gating system was selected for the front beam because the component height exceeds 800 mm and it is necessary to avoid excessive local overheating and provide smooth progressive filling. An open‑type gating system was adopted, with the following area ratios:
$$F_{\text{ladle}} : F_{\text{sprue}} : F_{\text{runner}} : F_{\text{ingate}} = 1 : 1.8 \sim 2.0 : 1.8 \sim 2.0 : 2.0 \sim 2.5$$
Based on these ratios and the ladle nozzle diameter of 100 mm, the diameters of the sprue, the runner and the ingates were determined as 140 mm, 140 mm and 100 mm, respectively. The three‑dimensional model of the initial gating system was established in UG and then meshed for flow simulation in ProCAST, as shown in Figure 3.
The initial gating system was simulated to evaluate the filling behaviour. The results indicated that the velocity of the liquid steel in the gating system reached 5 m/s in the early stage, which could cause mould erosion and sand inclusion in steel castings. More importantly, the filling process exhibited a phenomenon known as “misdirected flow” (乱引), whereby the upper ingates started to deliver the liquid metal before the mould cavity had reached that level. This behavior led to unwanted splashing, turbulence, and an increased risk of oxidation and air entrapment. The hydraulic analysis showed that the effective press head \(h_{\text{eff}}\) exceeded the distance between two adjacent layers of ingates, which was the primary cause of this misdirected flow.
9. Optimisation of the Gating System and Flow‑Field Verification
To eliminate the defects observed in the initial filling simulation, the following modifications were made to the gating system:
- The connecting runner was repositioned to align with the level of the first‑layer ingates, reducing the pressure head at the height of the auxiliary sprue.
- The cross‑sections of the second and third‑layer ingates were enlarged to a diameter of 120 mm to increase the flow rate at the upper levels.
- The pouring rate was controlled in stages: for the first 20 s, the flow rate was set to 120 kg/s, then it was increased to the full rate of 190 kg/s from 20 s to 30 s, and thereafter maintained at the full rate.
- The vertical distance between the top of the auxiliary sprue and the third‑layer ingates was reduced to increase the local pressure and promote flow into the upper levels.
The optimised filling simulation showed a significant improvement. The maximum velocity in the mould cavity was reduced to 2.79 m/s, which is acceptable for the prevention of mould erosion. The liquid metal entered the cavity layer by layer, starting from the bottom ingates and progressively switching to the upper ingates as the melt level rose. No misdirected flow or vortex was observed, leading to a calm and smooth filling of the mould cavity. This behaviour was also confirmed by the hydraulic analysis. A comparison between the initial and optimised gating hydraulic models is illustrated in the flow diagrams. The effective head of the initial process was higher than that of the optimised process, explaining the different filling behaviours.
| Parameter | Initial gating system | Optimised gating system |
|---|---|---|
| Maximum velocity in the cavity (m/s) | 5.0 | 2.79 |
| Misdirected flow | Yes | No |
| Filling stability | Poor | Good |
| Risk of oxidation and air entrapment | High | Low |
The optimised gating system also contributed to a better temperature distribution in the mould, since the upper layers of the casting were fed by more recent, hotter liquid metal, thus promoting direction solidification and improving the soundness of the steel castings.
10. Stress‑Field Simulation during Solidification of Large Steel Castings
Hot tearing and residual stress are major concerns in the production of large steel castings. In this study, a thermal‑elastoplastic model was used to simulate the stress evolution during solidification. The total strain rate can be decomposed into elastic, plastic and thermal components:
$$\mathrm{d}\varepsilon = \mathrm{d}\varepsilon_e + \mathrm{d}\varepsilon_p + \mathrm{d}\varepsilon_T$$
The thermal strain increment is expressed as:
$$\mathrm{d}\varepsilon_T = \alpha\, \mathrm{d}T + (T – T_0)\,\frac{\partial \alpha}{\partial T}\,\mathrm{d}T$$
where \(\alpha\) is the linear expansion coefficient, \(T\) is the instantaneous temperature, and \(T_0\) is the initial temperature. The von Mises yield criterion was adopted to judge the plastic deformation of the steel castings:
$$\bar{\sigma} = \frac{1}{\sqrt{2}} \sqrt{(\sigma_1-\sigma_2)^2 + (\sigma_2-\sigma_3)^2 + (\sigma_3-\sigma_1)^2}$$
If the von Mises equivalent stress exceeds the yield strength of the material at the corresponding temperature, plastic deformation occurs, and the possibility of hot tearing increases. A stress‑frame specimen was designed to study the influence of various factors on the stress evolution in steel castings. The specimen consists of a thick bar and a thin bar connected by two end bars, which mimics the typical sections encountered in the large front beam.
10.1 Analysis of the Stress‑Frame Tests
The temperature history of the stress‑frame specimen showed that the thin bar cooled faster than the thick bar in the early stage of solidification. Consequently, the thin bar contracted faster and was subjected to tensile stress, while the thick bar was subjected to compressive stress. At a certain critical time, the cooling rates of the two bars became equal, and thereafter the thick bar cooled faster, reversing the load on the two bars. This behaviour explains the origin of thermal stress in steel castings with non‑uniform sections. The maximum equivalent stress in the stress‑frame was found at the junction between the thin bar and the end bar, where stress concentration occurs. The hot‑tearing index was concentrated exactly in that region.
10.2 Parametric Studies
To identify the most relevant parameters that influence the stress field in steel castings, a series of parametric simulations were carried out using the stress‑frame model. The following factors were analysed.
| Mould material | Equivalent stress (MPa) | Hot tearing value (%) | Maximum deformation (cm) |
|---|---|---|---|
| Sodium silicate sand | 337.2 | 0.0338 | 0.1448 |
| Clay sand | 401.3 | 0.0403 | 0.1505 |
It was found that the sodium silicate sand has a lower high‑temperature strength and a better collapsibility than the clay sand, which reduces the constraint on the contraction of the steel castings and hence lowers the risk of hot tearing.
| Fillet condition | Equivalent stress (MPa) | Hot tearing value (%) | Maximum deformation (cm) |
|---|---|---|---|
| With fillets | 318.6 | 0.0317 | 0.1289 |
| Without fillets | 337.2 | 0.0338 | 0.1448 |
The presence of a fillet at the junction between bars markedly reduces the stress concentration and the tendency for hot tearing in steel castings.
| Condition | Hot tearing value (%) |
|---|---|
| Without chill | 0.0338 |
| With chill | 0.0215 |
The chills accelerate the cooling of the junction and increase the strength of the solidified shell, thus reducing the hot‑tearing tendency in steel castings.
| Pouring temperature (°C) | Equivalent stress (MPa) | Hot tearing value (%) | Maximum deformation (cm) |
|---|---|---|---|
| 1500 | 299.8 | 0.0270 | 0.1452 |
| 1530 | 307.3 | 0.0304 | 0.1440 |
| 1550 | 316.0 | 0.0322 | 0.1494 |
| 1580 | 337.2 | 0.0338 | 0.1448 |
| 1600 | 360.0 | 0.0402 | 0.1449 |
Higher pouring temperatures increase the equivalent stress and the hot‑tearing value in steel castings. Therefore, a lower pouring temperature is always recommended, provided the mould filling remains complete.
| Shakeout temperature (°C) | Equivalent stress (MPa) | Hot tearing value (%) | Maximum deformation (cm) |
|---|---|---|---|
| 400 | 780.3 | 0.0602 | 0.1945 |
| 450 | 707.0 | 0.0583 | 0.1868 |
| 500 | 614.8 | 0.0516 | 0.1796 |
| 550 | 520.9 | 0.0437 | 0.1705 |
| 600 | 458.5 | 0.0413 | 0.1612 |
| 650 | 412.7 | 0.0397 | 0.1537 |
| 700 | 337.2 | 0.0338 | 0.1448 |
The shakeout temperature strongly affects the residual stress of steel castings. A higher shakeout temperature reduces the final residual stress because the casting is allowed to cool more uniformly inside the mould. However, too high a shakeout temperature may result in an insufficient strength of the casting during handling and excessive oxidation. Therefore, an optimal value must be chosen for each component.
11. Application to the Extruder Front Beam
Based on the findings from the stress‑frame experiments, the initial casting process of the front beam was assessed. The initial process used clay sand for both the mould and the core, and the pouring temperature was 1580 °C. The stress‑field simulation of the initial process revealed several critical locations with equivalent stresses exceeding the yield limit of the material, including the junction between the gating system and the casting, the area around the tension hole, and some sharp corners. These areas displayed a high risk of hot tearing in the steel castings.
To improve the stress state, the following modifications were made to the casting process:
- The clay sand mould was replaced by a sodium silicate sand mould with better collapsibility.
- The clay sand core was replaced by an ester‑hardened sodium silicate sand core with a loose layer to improve the deformability of the core during contraction of the steel castings.
- The fillet radius at the sharp corners of the casting was increased from 30 mm to 50 mm, and a radius of 10 mm was added at the junction between the gating system and the casting.
- The shakeout temperature was increased to 550 °C, allowing the casting to cool more uniformly in the mould before the restraint is removed.
- The pouring temperature was reduced from 1580 °C to 1550 °C.
- A methanol‑based refractory coating was applied to both the mould and the core surfaces and dried, which reduces the friction between the casting and the mould and prevents sand burning.
After the modification, the stress‑field simulation of the optimised process was performed. The results showed that the equivalent stress values in all critical regions remained below the yield strength of the material, confirming that hot tearing would not occur in the casting body. The thermal‑crack prediction showed no hot tearing within the front beam, which was fully consistent with the equivalent‑stress analysis. The deformation analysis of the optimised casting indicated a maximum displacement of approximately 4.4 cm at the surface of the casting, which is acceptable for the required machining allowance of large steel castings of this size.
12. Conclusion and Perspectives
In this study, a comprehensive numerical simulation of the casting process for a large extruder front beam was carried out using the commercial finite element software ProCAST. The evolution of the temperature field, the flow field during mould filling, and the stress field during solidification were systematically investigated. The main findings and conclusions can be summarised as follows:
- The temperature‑field simulation predicted the locations of shrinkage porosity in the large steel castings and proved that a combination of insulated risers, exothermic sleeves, padding and external chills can successfully establish directional solidification. The shrinkage defects were confined to the risers, resulting in a fully sound casting body. The yield rate of the casting was increased from 59.79% to 69.67% while maintaining the casting quality.
- The flow‑field simulation of the filling process demonstrated the importance of designing a stepped gating system for the large steel castings. The misdirected flow induced by an improper pressure head was correctly predicted and subsequently eliminated by a repositioning of the connecting runner, an appropriate enlargement of the upper ingate section and a staged pouring rate control.
- The stress‑field simulation identified hot tearing as the critical defect mode for the large steel castings. The parametric study on the stress‑frame specimen revealed that high‑temperature collapsibility of the mould, fillet radius, cooling chills, pouring temperature and shakeout temperature all affect the stress evolution. An improved process, including sodium silicate sand, larger fillets, and a higher shakeout temperature, successfully eliminated the risk of hot tearing and reduced deformation of the steel castings.
The study confirmed that the numerical simulation of the casting process, when rigorously coupled with hydraulic calculations and defect‑prediction criteria, provides an efficient and reliable tool for the design and optimisation of the casting of large steel castings. The methodology established in this study can be further extended to the production of steel castings with similar dimensions and service requirements.