In the present work, I focus on a heat-resistant steel casting of a gas turbine compressor support ring that was newly introduced in an enterprise. Compared with ordinary castings, this type of casting is larger in size, more complex in structure, consumes a large amount of molten metal, and contains regions with substantial wall thickness. During the mold filling and solidification processes, it becomes more difficult to establish appropriate temperature gradients and maintain a desired solidification sequence. Quality problems such as misruns, shrinkage porosity, shrinkage cavities, and hot tearing deformation are therefore more likely to occur. In actual production, stringent requirements are placed on mold structure, gating system design, and temperature control. To improve the overall casting quality of the compressor support ring, it is essential to employ the casting simulation software ProCAST to simulate the casting process, identify possible defects in the casting, and optimize the casting process plan accordingly.
Combining the theoretical knowledge of gravity sand casting with the structural dimensions of the compressor support ring casting, I compared the advantages and disadvantages of various gating systems and determined that a bottom gating system was the most suitable choice. In addition, three feeding risers and feeding pads were installed at thick-walled sections of the casting to ensure that the casting could receive sufficient molten metal for feeding during solidification. The three-dimensional model of the compressor support ring casting was constructed using NX11.0, and finite element mesh generation as well as pre-processing parameter settings were completed with the assistance of Visual-Mesh. I subsequently performed a simulation calculation of the mold filling and solidification processes for the initial scheme. Through analysis of the results, I found that in the early stage of mold filling with the original scheme, the initial flow velocity of the molten metal was excessively high, which caused a certain degree of entrained gas and oxidation in the lower part of the casting. The solidification process did not establish an effective temperature gradient distribution, resulting in three shrinkage cavities and two shrinkage porosity defects inside the casting.
To address the problems identified in the original scheme, I redesigned the feeding system. The ordinary sand risers were replaced with exothermic insulating risers, and a chilling system was newly introduced by placing two pairs of cold irons on both sides of the middle riser. Using ProCAST software, I simulated the temperature field during the solidification process of the new process plan. The cold irons provided an effective chilling effect, and the shrinkage porosity on both sides of the riser in the middle of the casting was eliminated. The exothermic insulating risers enabled the formation of a bottom-to-top directional solidification sequence in the riser area, thereby ensuring effective feeding and preventing internal shrinkage defects. This redesign significantly improved the casting quality and resolved the defects present in the original scheme. It not only enhanced the casting process but also provided valuable technical support and assurance for subsequent production practices.
Based on the new scheme, I designed orthogonal experiments to further optimize the casting pouring process parameters. Through direct analysis of the orthogonal experimental design results, I determined the optimal parameter combination for the pouring process: a pouring temperature of 1575 °C, a pouring speed of 100 kg/s, and a sand mold temperature of 20 °C. Simulation analysis was conducted using this optimized parameter combination, and no shrinkage porosity or shrinkage defects were found. Finally, the simulation results were validated through actual production. Non-destructive testing techniques were employed to inspect the castings, and no macroscopic defects were found in the castings. This demonstrates the effectiveness of the optimized process scheme and pouring parameters for the compressor support ring, significantly improving the casting quality.
Introduction
The research presented in this thesis originates from a key scientific and technological project undertaken by a domestic foundry enterprise. This project primarily focuses on defects that occur during the mold filling and solidification processes of large steel castings used in gas turbines and steam turbines. Due to structural constraints of certain components, some regions cannot be completely and fully fed, ultimately resulting in shrinkage cavities and shrinkage porosity. Alternatively, during the pouring process, secondary oxidation inclusions and entrained gases may be generated, which cannot effectively float up to the riser in the later stages, ultimately leading to inclusions and gas porosity defects. Abrupt wall thickness transitions or hindered free contraction may also produce high-stress zones, ultimately leading to crack defects. To avoid or reduce defects during the casting process, enterprises urgently need to optimize their casting processes.
Castings, as a metal hot-working technology with a long history, have always occupied an important position in the manufacturing industry. At the same time, the casting process is one of the most energy-intensive manufacturing processes. According to historical records and unearthed artifacts, as early as five thousand years ago, humans had already mastered basic casting techniques and cast simple copper axes. China is one of the ancient civilizations in the world that mastered the art of casting, with a casting history of more than 5000 years. Casting is a forming method in which metals or alloys are melted at high temperatures and, under atmospheric, specially protected gas, or vacuum environments, are filled into pre-prepared mold cavities through the action of external fields alone or in combination, such as gravity, pressure, centrifugal force, or electromagnetic force, and then solidified into castings with a certain shape, size, and properties. Casting is generally divided into sand casting and special casting processes. Sand casting is one of the oldest technologies for producing metal parts. Steel, iron, and most non-ferrous alloy castings can be produced by sand casting. The sand molds used in the casting process are low in cost and suitable for the mass production of large castings. Special casting processes are subdivided into investment casting, pressure die casting, centrifugal casting, lost foam casting, continuous casting, vacuum casting, and permanent mold casting. Compared with sand casting, the castings obtained by special casting processes have better surface finish, dimensional accuracy, and mechanical properties.
The large gas turbine valve body casting is a high-tech product that integrates material development, melting, pouring, heat treatment, machining, and inspection. It is one of the key fundamental components in the manufacturing of heavy-duty gas turbines. Data show that global casting production has fluctuated around 100 million tons in recent years. In 2020, affected by the COVID-19 pandemic, global casting production fell by 2.75% year-on-year. In 2022, global casting production reached 110 million tons. Since 2000, when China surpassed the United States to become the world’s largest casting producer, domestic casting output has generally shown a growth trend. Since 2011, the growth rate has transformed from “high-speed growth” to “medium-to-low speed growth.” In recent years, due to the stable operation of the domestic economy and a favorable international environment, China’s foundry machinery industry has maintained a high growth trend. In 2022, China’s casting output reached 56.95 million tons, a year-on-year increase of 5.4%. Among this, castings for power generation equipment and the electric power field completed 2.48 million tons. On the other hand, in recent years, investment in foundry equipment by large-scale casting enterprises in China has increased significantly. Advanced equipment such as 3D printers, automated molding lines, automatic pouring machines, industrial robots, and large die-casting machines has been gradually applied in casting production. Statistics show that from 2011 to 2019, the number of newly added molding lines in China was as high as more than 5,100. This not only reflects the rapid improvement in China’s foundry equipment level but also shows that the overall equipment level of a few enterprises has approached or even reached the advanced level of industrialized countries. Furthermore, with the optimization of China’s economic structure and the adjustment of industrial layout, the casting industry is gradually shifting to the central and western regions. This trend has not only accelerated the economic development of the central and western regions but also promoted the rapid development of casting industry clusters and base construction. These clusters and bases have not only improved the concentration of the casting industry but also promoted technological innovation and industrial upgrading, laying a solid foundation for the sustainable development of China’s casting industry.
China’s energy structure, characterized by high coal use and low oil use, determines that thermal power will remain the main source of electricity generation in China for some time to come. With China’s goal of peaking carbon dioxide emissions before 2030 and striving to achieve carbon neutrality by 2060, the national policy environment that vigorously promotes industrial structural adjustment, optimization and upgrading, energy conservation and emission reduction, and further eliminates small thermal power units has been established. In the future, thermal power will develop toward large capacity, high parameters, energy conservation, and environmental friendliness. According to the Rankine cycle efficiency formula, the most effective way to improve cycle efficiency is to increase the turbine output heat, which means adopting higher gas parameters, increasing the boiler service temperature and gas pressure. Heavy-duty gas turbines, by raising temperature and pressure, have achieved combined cycle power plant efficiencies as high as 60%-61%, improving coal combustion and power generation efficiency while reducing exhaust emissions. The gas-steam combined cycle burning natural gas is one of the most efficient clean power generation technologies that mankind has mastered. At present, China’s research and manufacturing of large high-power heavy-duty gas turbines have reached an advanced level in the world.
At present, components such as the compressor, high-medium-low pressure combustion chamber valve shells, main steam valve covers, and seat frames of gas turbines used in thermal power plants are generally manufactured by sand casting. These large steel castings are large in size and wall thickness, and the amount of molten steel required for pouring generally ranges from several hundred kilograms to several tons or more. During the mold filling and solidification processes of steel castings, casting defects such as slag inclusions, gas porosity, shrinkage cavities, and cracks often occur due to various causes, which can lead to rejection. In the field of power generation equipment, castings with various casting defects account for approximately 10%-15% of the annual output of steel castings, which is commonly referred to as the rejection rate of 10%-15%. This results in a high rejection rate and a short service life. If casting processes are not optimized, 250,000 to 400,000 tons of steel castings are scrapped each year, with an annual loss of more than several hundred million yuan. If trial-and-error methods are employed for experimentation, the manufacturing cycle is long, the research and development cost is high, and the production difficulty is great. The use of numerical simulation technology to simulate the mold filling and solidification processes of casting, through multi-physics coupling simulation, can predict the size and location of defects that may occur during the casting process. This eliminates a large number of preliminary casting experiments, shortens the product development cycle, and saves material and labor costs.
In this context, combined with the casting problems of a compressor support ring large steel casting in a mechanical foundry, I optimized its casting process to address the problems caused by the formation of numerous defects during the sand casting process that led to an excessively low casting yield and the need for extensive subsequent machining to repair the defects. The goal was to obtain large steel castings with good mold filling and high quality, promote the development of the enterprise in the direction of high efficiency and high quality, and create higher economic benefits for the enterprise.
Basic Principles of Casting and System Design
The common process flow for sand casting of large steel castings adopted by casting enterprises is: “making wood patterns – molding – melting – pouring – shakeout – removing gates and risers for cleaning – inspection and warehousing.” In the wood pattern making stage, in order to improve production efficiency, enterprises often use automatic equipment such as CNC machine tools to process wood patterns for casting. First, three-dimensional modeling software such as UG or Pro-e is used to draw the original three-dimensional model, and then a wood pattern drawing is prepared on the basis of this model, along with the corresponding CNC programming software. The wood pattern drawing is imported into the software, and data conversion and optimization are carried out to generate a program that can be used for CNC machine tool processing. Then, wood materials suitable for CNC machining are selected, usually pine wood with fine and dense structure, easy processing, and relatively low cost. Before processing, the operating status of the CNC machine tool is checked, and appropriate tools and cutting parameters are selected according to the processing requirements. The prepared program is imported into the CNC machine tool for processing. During the processing, it is necessary to continuously monitor the machining quality and progress, and timely adjust parameters or replace tools. After processing is completed, the wood pattern is removed from the CNC machine tool and subjected to quality inspection. Fine adjustment or repair of details is carried out manually if necessary.
Sand mixing is an important process in casting production. Raw materials such as quartz sand or phenolic resin sand are thoroughly mixed with binders, water solvents, additives, and other raw materials in an automatic sand mixing device in certain proportions. Combined with the wood pattern, sand molding and core making are completed to obtain sand molds and sand cores, and the two are then closed in a flask to obtain the mold. At the same time as sand mixing and molding, the melting workshop simultaneously prepares charge materials. An appropriate melting furnace is selected for metal melting to obtain liquid metal that meets the temperature and composition requirements. The molten steel melted in the refining furnace is injected into the mold with a ladle, allowing the molten steel to fill the entire mold cavity. Attention should be paid to the pouring speed during the pouring process to prevent splashing of molten steel. After pouring, sufficient cooling time is required to allow full solidification of the mold. When the casting cools to a specific temperature range, the shakeout process is carried out. During this process, the flask and the casting are first separated, and the casting is then carefully cleaned, including removing gates and risers, residual core sand, and removing sand adhesion, flash, burrs, and oxide scale on the surface of the casting. Finally, after strict quality inspection, defects found are repaired, and subsequent processes such as heat treatment and painting are carried out in sequence to ensure that a qualified casting is finally obtained.
The gating system is an important part of the entire casting system. As a channel for introducing liquid metal into the mold, it can reasonably control the flow velocity and mold filling time of the liquid metal during the filling process, ensure that the metal liquid smoothly fills the entire mold cavity, and provide appropriate feeding during the solidification process of the casting. The gating system is usually composed of a pouring cup and runners, and the runners include sprue, cross runner, and ingate. The rationality of gating system design is directly related to the quality of the casting. If the design of the gating system is inadequate, various defects such as gas porosity, inclusions, cracks, sand holes, and shrinkage porosity may occur in the casting. According to the daily production records of enterprises, about 20%-30% of castings requiring later repair in the casting process are caused by unreasonable gating system design. The design of the gating system mainly includes the selection of the structural type of each component, the calculation of section ratios, the location of introduction, and the method of introduction.
In general, the design of the gating system needs to follow the following two principles: The gating system should ensure the smooth progress of the pouring process and also create the necessary conditions for feeding during mold filling. The structural design of the entire gating system should be as simple as possible, and the volume should be controlled within an appropriate range to reduce resource consumption and improve production efficiency. Specifically, the following five points are included:
1) Determine the appropriate type of gating system so that it can achieve the best pouring effect.
2) Fill the casting within a certain pouring time to avoid defects such as cold shuts, oxidation inclusions, misruns, and shrinkage porosity.
3) Control the speed and direction of molten metal entering the mold so that the molten metal flows steadily, preventing fluid turbulence and gas entrainment that lead to excessive oxidation of the molten metal. Also avoid splashing and impact on the mold and core due to excessive flow velocity, which would result in poor casting surface quality.
4) Effectively control the temperature gradient distribution during the solidification process to reduce or avoid cracks, shrinkage porosity, and shrinkage cavities in the casting.
5) Control the volume and structural complexity of the gating system as much as possible to facilitate molding, reduce the workload of later cleaning processes, and reduce metal liquid consumption.
Gating systems can be classified by the position at which molten metal is introduced into the mold cavity. A top gating system introduces molten metal from the top of the mold cavity. This system is characterized by high filling speed, short filling time, and strong filling capacity. During the mold filling and subsequent solidification process, the temperature distribution of the casting exhibits a pattern of high temperature at the top and low temperature at the bottom. This temperature gradient helps the casting achieve directional solidification from the bottom upward. In addition, this feature effectively enhances the feeding effect of the top riser during the solidification process of the casting. However, the molten metal falls from a height in the mold cavity, causing severe impact on the mold, and phenomena such as splashing, gas entrainment, and aspiration are prone to occur, leading to defects such as gas porosity and oxidation inclusions. In the meantime, the exhausting and slag-intercepting effects are poor. This system is often used for small aluminum alloy castings. In the bottom gating system, the ingate is placed at the bottom of the casting, and the molten metal is injected into the mold cavity from the bottom. The filling process of this system is stable, and it is not easy to impact the mold, splash, or entrain gas. During the entire mold filling process, the cross runner is basically maintained in a full state. This feature not only helps to intercept slag but also has superior exhausting effects. However, during solidification, the temperature distribution shows a trend of high temperature at the bottom and low temperature at the top, which is not conducive to directional solidification from the bottom upward and weakens the feeding effect of the top riser. In addition, the molten metal enters the mold cavity from the bottom, causing overheating at the ingate and the bottom of the casting, which may lead to coarse grains and defects such as shrinkage porosity. The middle gating system introduces molten metal at a position between the top and bottom gating systems. It combines the advantages and disadvantages of both. This filling method combines the strong mold filling capacity of the top gating system and the stable filling process of the bottom gating system. It is particularly suitable for small and medium-sized castings with low height and large horizontal dimensions. In the step gating system, multiple ingates are set at different heights of the casting to ensure that the molten metal fills the mold layer by layer from bottom to top. The filling process is stable, and exhaust gases are removed smoothly. Before the solidification process begins, the temperature of the upper part of the casting is higher than that of the lower part, forming a temperature gradient distribution from bottom to top, which is beneficial for directional solidification and riser feeding. Multiple ingates can also alleviate the overheating problem at the ingate position. However, this gating system is more complex in structure and affects the subsequent cleaning process. It is prone to “chaotic pouring” conditions, which may result in undesirable temperature distribution in the casting.
Gating systems can also be classified by the cross-sectional area relationship of their components. The shrinking gating system satisfies the condition that the sum of sprue sections is greater than that of the cross runner sections, which is greater than that of the ingate sections. The sprue, cross runner, and ingate sections decrease in sequence. This system has good filling performance, and the ingate section is the flow-control section. At the initial stage of pouring, the gating system can be quickly filled, which is favorable for slag floating and interception in the cross runner. However, the molten metal flows at an accelerating speed, causing severe impact on the mold, splashing, and severe oxidation. This type of gating system is small in volume, consumes less metal, and is easy to clean later. The expanding gating system satisfies the condition that the sum of sprue sections is less than that of the cross runner sections, which is less than that of the ingate sections. The advantages and disadvantages of this system are opposite to those of the shrinking gating system. The flow velocity becomes lower and lower, the filling is stable, and impact oxidation can be reduced. The sprue section is the flow-control section. At the early stage, the ingate and cross runner are in a non-pressure state and have poor slag interception effects, which can be improved by placing a filter screen. This gating system is large in volume, consumes more metal, and is not easy to clean later. The semi-expanding gating system satisfies the condition that the sum of sprue sections is less than that of the cross runner sections, but greater than that of the ingate sections, and the sum of sprue sections is greater than that of the ingate sections. The cross runner section is the largest, and the ingate section is the smallest. The ingate section is the flow-control section. Its characteristics are between those of the shrinking type and the expanding type. The filling speed is moderate, and the filling and slag-intercepting capabilities are relatively good.
Riser design is of primary importance in the feeding system. During solidification, the casting volume contracts due to the thermal expansion and contraction effect. If sufficient compensation cannot be provided, shrinkage porosity and shrinkage cavities are likely to appear in the alloy. Therefore, it is necessary to supplement the molten metal in a timely manner. A riser is the cavity in the mold used to store supplementary molten metal. Reasonable riser design can effectively avoid or reduce shrinkage defects in the casting. Riser design should follow the following four requirements: The solidification of the riser should be later than or simultaneous with that of the hot spot. The size of the riser should ensure that sufficient molten metal is available to compensate for the liquid shrinkage and solidification shrinkage of the casting to avoid leaving shrinkage cavities in the casting. During the solidification process, a certain feeding pressure and feeding channel must be maintained. For castings with a wide crystallization temperature range and propensity to develop dispersed shrinkage porosity, it is also necessary to combine the gating system, cold irons, and feeding pads to achieve directional solidification from the bottom of the casting to the top of the riser. In addition to satisfying the above three requirements, the riser volume should be minimized, the structure should be simple, and subsequent cutting and cleaning should be convenient.
In actual production, foundries often classify risers according to the heating method. They are usually divided into ordinary sand risers, insulating risers, exothermic risers, exothermic insulating risers, oxygen-enhanced risers, arc-heated risers, gas-heated risers, and so on. The feeding effect of the riser is not only related to the size of the riser and the solidification time of the casting but also depends on whether the feeding channel is unobstructed during the solidification process of the casting. When a clear feeding channel is maintained between the riser and the fed part of the casting, the molten metal in the riser will continuously compensate for the volume contraction of the casting. The larger the expansion angle of the feeding channel, the smoother the channel and the easier the feeding. On the other hand, the effective feeding distance of the riser also affects the feeding effect. The effective feeding range of the riser is equal to the effective feeding distance of the riser plus the riser radius. The number of risers can be determined by the effective feeding range of the riser.
The modulus method is a widely used approach for riser design of steel castings and is recognized as a convenient and practical method. The modulus method should satisfy two requirements. First, the solidification time of the riser must not be less than the solidification time of the part of the casting to be fed. When using the modulus concept for riser calculation, when the riser modulus is greater than or equal to the casting modulus, the riser solidification will be later than the casting. Usually, for steel castings, the modulus satisfies the following relationship to achieve feeding. For top open risers:
$$M_r = (1.1 \sim 1.2) M_c \tag{1}$$
For side blind risers:
$$M_c : M_n : M_r = 1 : 1.1 : 1.2 \tag{2}$$
When pouring through the riser:
$$M_c : M_n : M_r = 1 : (1 \sim 1.03) : 1.2 \tag{3}$$
The riser neck length:
$$L = 2.4 M_c = 2 M_r \tag{4}$$
Second, the riser must have sufficient molten metal to compensate for the liquid shrinkage and solidification shrinkage of the casting and the riser, to avoid leaving shrinkage cavities in the casting.
$$V_r – V_{rf} = \beta (V_r + V_c) \tag{5}$$
To ensure that no shrinkage cavities appear in the casting, equation (5) can also be rewritten as:
$$\beta (V_r + V_c) \leq V_r \eta \tag{6}$$
where the feeding efficiency is:
$$\eta = \frac{\beta (V_r + V_c)}{V_r} = \frac{V_r – V_{rf}}{V_r} \times 100\% \tag{7}$$
The typical feeding efficiencies of various riser types are listed in Table 1.
| Riser type | Cylindrical / waist cylindrical | Spherical | Repouring | Exothermic insulating | Atmospheric pressure | Compressed air | Gas bomb |
|---|---|---|---|---|---|---|---|
| η (%) | 12-15 | 15-20 | 15-20 | 25-30 | 15-20 | 35-40 | 30-35 |
Exothermic insulating risers involve placing an insulating sleeve made of insulating material in the riser cavity and covering the top of the riser with an insulating board. This can prolong the solidification time of the riser and greatly improve the feeding efficiency of the riser. The insulating material reduces the heat dissipation of the riser, thereby reducing the effective cooling surface of the riser. In the riser equation, the riser modulus becomes:
$$M’ = \frac{V}{aS_{side} + bS_{top}} = \frac{M}{a + (b-a)S_{top}/(S_{side}+S_{top})} \tag{8}$$
For blind insulating risers, a = 0.7 and b = 0.7, giving:
$$M’ = \frac{M}{0.7} = 1.43M \tag{9}$$
For open insulating risers, the height-to-diameter ratio is often taken as 1.3, so:
$$\frac{S_{side}}{S_{top}} \approx 5.2 \tag{10}$$
$$M’ = \frac{M}{0.7 + 0.3/(5.2+1)} = 1.33M \tag{11}$$
During the casting process, a chilling material is usually placed inside or on the surface of the mold cavity to adjust the local cooling rate of the casting. This chilling material is called a cold iron. The main functions of the cold iron are to expand the effective feeding distance of the riser, accelerate the cooling rate of hot spots in the casting, and improve the matrix structure performance of certain special parts of the casting. Cold irons are placed at locations where accelerated cooling is required, such as thick sections and hot spots, where defects such as shrinkage cavities and shrinkage porosity are prone to occur. For large castings, cold irons can be placed at the bottom and side surfaces of the casting, because these parts are less affected by the heat of the top. The thickness of the cold iron is one of the key parameters that affect the cooling rate, solidification sequence, and internal quality of the casting. In actual production, the thickness of the cold iron is determined according to the function of the cold iron and the size of the hot spot at the location where the cold iron is placed. For steel castings, the cold iron thickness is typically 0.3 to 0.8 times the thickness of the hot section. For aluminum-magnesium alloy castings, it is 0.8 to 1.9 times the hot spot thickness, and for gray iron castings, 0.25 to 0.5 times. Whether to use single-sided or double-sided cold irons is determined by the empirical formula between the hot spot thickness t and the connected wall thickness T:
When t ≤ 2T, a single-sided cold iron is used. When 2T ≤ t ≤ (3~4)T, double-sided cold irons are used. When t > 4T, cold irons alone cannot satisfy the requirement of eliminating shrinkage porosity, and they must be used in combination with risers to achieve the best effect.
The cooling area of the cold iron is its working surface area. For the thick-walled flat surfaces of large steel castings, a group of small cold irons is usually scattered to form a cold iron group. The effect of the cold iron on the temperature field mainly depends on the size of the chilling surface area of the cold iron. With the increase of the cold iron surface area, the effective modulus of the casting decreases. The equivalent surface area of the casting is the sum of the surface area of the part without cold irons and the working surface area of the cold irons:
$$S_1 = S_n + yS_y = S_0 + (y-1)S_y \tag{12}$$
From which the working surface area of the cold iron is derived as:
$$S_y = \frac{S_0 – S_1}{y-1} \tag{13}$$
Let:
$$M_0 = \frac{V_0}{S_0} \tag{14}$$
$$M_1 = \frac{V_0}{S_1} \tag{15}$$
Substituting equations (14) and (15) into (13):
$$S_y = \frac{V_0 (M_0 – M_1)}{(y-1)M_0 M_1} \tag{16}$$
where S1 is the equivalent surface area of the casting, Sn is the surface area of the part of the casting where cold irons are placed, y is the chilling amplification coefficient, Sy is the surface area of the cold irons, S0 is the geometric surface area of the casting, M0 is the original modulus of the casting, and M1 is the equivalent modulus of the casting after using cold irons.
In general, the cold iron plays an important role in controlling the solidification process, preventing defects, and improving casting quality. Selecting the appropriate cold iron material, designing cold iron shapes and dimensions that are convenient to use and provide good chilling effects according to the casting structure, and choosing a reasonable cold iron layout can achieve the chilling effect on the hot spots of the casting, thereby reducing or eliminating casting shrinkage porosity defects.
Numerical Simulation Software and Theoretical Basis
In this chapter, I introduce the casting CAE software used in this research, namely ProCAST, and present the numerical simulation theoretical basis, including the numerical models of the mold filling process and the solidification process, the treatment of boundary conditions, and the methods for defect prediction.
ProCAST software was originally developed and released by UES Corporation in the United States, and was subsequently acquired by the French software company ESI Group in 2002, becoming a subsystem of its Visual-Environment platform. Its principle is based on the finite element method and computational fluid dynamics, and it can discretize and solve physical fields in the casting process, including temperature fields, stress fields, flow fields, and so on. The software is widely used for the simulation of the casting process. ProCAST can be used to track the complete casting process: mold filling, cooling and crystallization of the alloy, formation of defects, growth of residual stresses, deformation of the casting, etc. All the steps for solving the problem, such as inputting the geometry of the casting, constructing the finite element mesh, specifying the process parameters, and analyzing the results, are executed in Visual-Cast. ProCAST discretizes the casting process into a series of grid cells and establishes equations within each grid cell. Based on the physical parameters in the equations and the boundary conditions between adjacent cells, the numerical solutions of each physical field in each cell are obtained. At the same time, ProCAST takes into account the influence of material properties changing with temperature and stress, including the thermal and physical properties of materials, elastic modulus, plastic behavior, and other aspects. In addition to being applied in the field of sand casting, ProCAST also provides simulation for various casting processes such as high-pressure die casting, low-pressure die casting, gravity casting, centrifugal casting, investment casting, and continuous casting. Therefore, it is regarded by the industry as providing one of the most powerful, comprehensive, and accurate casting simulation solutions.
The process of completing a casting simulation using ProCAST requires different sub-programs to successively complete the corresponding stages. The main sub-programs used are: Visual-Mesh module, the pre-processing finite element mesh generation module of ProCAST, which can automatically generate finite element meshes. It can directly import modeling results from UG, CAD, Pro-E, and other software, making mesh generation more convenient and faster. The Visual-Cast module is mainly responsible for pre-processing settings related to casting calculation parameters. The DataCast/ProCast module is the casting calculation solver, mainly responsible for processing operations. The Visual-Viewer module is responsible for displaying the results of casting simulation calculations, providing a more intuitive view and analysis of the simulation results.
The typical process flow of ProCAST sand casting simulation includes the following steps: First, create three-dimensional models using three-dimensional modeling software such as UG, CAD, or Pro-E to create the casting body, gating system, risers, cold irons, and other models, and save these models in a file format acceptable to Visual-Mesh (such as iges, step, stl). Second, mesh division and model checking: use the Visual-Mesh module to perform mesh division on the input three-dimensional model files, generate tetrahedral meshes, check and repair the surface mesh and volume mesh models, and finally generate a volume mesh .mesh file containing node numbers, element numbers, material numbers, and other information. Third, parameter settings: in the Visual-Cast module, assign material properties to the volume mesh model, set interface conditions, boundary conditions, initial conditions, and simulation parameters, and through parameter self-checking, determine and correct erroneous parameters. Fourth, simulation calculation and solving: use the Visual-Cast module to perform the simulation analysis calculation of the casting process, including mold filling process and solidification process calculations. Fifth, results display and analysis: generate visual simulation results in the Visual-Viewer module, where the flow field and temperature field of the mold filling process, the solidification sequence and solidification time of the solidification process, and the temperature curves of specific points can be intuitively observed. Combined with relevant defect criteria, the defects that may occur during the casting process are predicted, providing the basis for subsequent process improvement.
The numerical simulation of the casting process involves multiple disciplines, mainly including three-dimensional modeling, fluid mechanics, heat transfer, solidification theory, numerical analysis, and physical chemistry. The purpose of casting numerical simulation is to make the liquid metal forming process visible, so as to gain a deeper understanding of the casting process, optimize the casting process, and improve casting quality. The casting process is mainly divided into two stages: mold filling and solidification. The filling and solidification of the casting is a non-steady-state process involving complex factors such as high temperature, variable melt state, substantial heat dissipation and transfer, and phase changes. At the macroscopic level, three physical phenomena are involved: liquid flow, cooling, and alloy contraction. Therefore, the establishment of the numerical simulation calculation model of casting needs to consider these factors comprehensively.
In casting numerical simulation, the mold filling process is a key part. The main research focus is on the dynamic process of liquid metal entering the mold through specific channels and filling the entire mold cavity, as well as the design and coordination of the liquid metal with the gating system. This process involves multiple aspects such as the flow behavior of the liquid metal, the interaction with the mold, and possible defects. In addition to the mold filling process, the solidification process is also an important part of casting numerical simulation. The solidification process involves heat transfer, phase transition, and changes in the microstructure. Heat transfer plays a crucial role in the solidification process. Its characteristics can be summarized by the “one heat, two migrations, three transfers” aspects. “One heat” indicates that heat transfer occupies a core position in the solidification process and is the key factor driving the entire solidification process. “Two migrations” involves the migration phenomena of two key interfaces during metal solidification, namely the solid-liquid interface and the metal-mold interface. “Three transfers” reveals the complexity of the metal solidification process. It is a three-dimensional heat transfer process in which momentum transfer, mass transfer, and heat transfer are closely coupled, jointly affecting the progress and results of solidification. To accurately simulate the solidification process, comprehensive consideration must be given to factors such as the thermal properties of materials, latent heat of solidification, heat conduction conditions, and external cooling conditions.
The mold filling process of the casting refers to the process from the injection of molten metal from the ladle into the gating system until the mold cavity is completely filled. In actual production, the mold filling process of the molten metal is a turbulent flow and heat transfer process. Since turbulent models are much more complex than laminar models, most commercial casting software simplifies the mold filling process to a laminar flow and heat transfer model. The numerical simulation methods of the mold filling process mainly include the SOLA-VOF method, SIMPLE method, MAC and SMAC methods, DFDM method, and the conservative scalar method. During the mold filling stage, the core feature is the free flow of high-temperature molten metal in the complex mold cavity. During this process, the molten metal is not only obstructed by the mold cavity walls but also exchanges heat with the mold cavity and the environment. To accurately describe this complex dynamic process, the molten metal is usually assumed to be an incompressible, viscous fluid with constant density, and the influence of turbulent flow is neglected. On this basis, the flow and heat transfer mechanisms of the molten metal must strictly follow the three basic conservation laws: the continuity equation, the momentum equation, and the energy equation.
The continuity equation describes the mass conservation of the molten metal during the mold filling process, i.e., the difference between the mass entering and leaving the control volume equals the change in mass within the control volume. Its mathematical expression is:
$$\frac{\partial \rho}{\partial t} +
abla (\rho \mathbf{V}) = 0 \tag{17}$$
Since the molten metal is generally treated as an incompressible fluid, its density is independent of time and position, and equation (17) can be rewritten as:
$$\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y} + \frac{\partial w}{\partial z} = 0 \tag{18}$$
where ρ is the fluid density, t is the flow time, ∇ is the vector differential operator, V is the flow velocity vector, and u, v, w are the components of the flow velocity in the x, y, and z directions.
The momentum conservation equation, also known as the Navier-Stokes equation, describes the momentum conservation of the molten metal during the mold filling process, i.e., the resultant force of gravity, pressure, surface tension, etc., acting on the molten metal equals its rate of momentum change. For an incompressible liquid, this equation can be expressed in the three coordinate directions as:
$$\rho \left( \frac{\partial u}{\partial t} + u\frac{\partial u}{\partial x} + v\frac{\partial u}{\partial y} + w\frac{\partial u}{\partial z} \right) = -\frac{\partial P}{\partial x} + \rho g_x + \mu
abla^2 u \tag{19}$$
$$\rho \left( \frac{\partial v}{\partial t} + u\frac{\partial v}{\partial x} + v\frac{\partial v}{\partial y} + w\frac{\partial v}{\partial z} \right) = -\frac{\partial P}{\partial y} + \rho g_y + \mu
abla^2 v \tag{20}$$
$$\rho \left( \frac{\partial w}{\partial t} + u\frac{\partial w}{\partial x} + v\frac{\partial w}{\partial y} + w\frac{\partial w}{\partial z} \right) = -\frac{\partial P}{\partial z} + \rho g_z + \mu
abla^2 w \tag{21}$$
The energy conservation equation describes the heat conservation of the molten metal during the mold filling process, including the heat exchange between the molten metal and the mold, the heat conduction inside the molten metal, and the heat generated by viscous friction during the flow of the molten metal:
$$\rho c \left( \frac{\partial T}{\partial t} + u\frac{\partial T}{\partial x} + v\frac{\partial T}{\partial y} + w\frac{\partial T}{\partial z} \right) = \frac{\partial}{\partial x}\left( k\frac{\partial T}{\partial x} \right) + \frac{\partial}{\partial y}\left( k\frac{\partial T}{\partial y} \right) + \frac{\partial}{\partial z}\left( k\frac{\partial T}{\partial z} \right) + S \tag{22}$$
The numerical theory of the solidification process focuses on constructing the heat transfer model of solidification, using numerical methods to obtain the solidification sequence of the casting, determine the last solidified region, and thus predict possible defects. During the solidification process, the high-temperature molten metal gradually releases heat to the mold and the surrounding environment, gradually cooling and solidifying, and finally obtaining a qualified casting. The solidification process of the casting goes through three stages: liquid state, liquid-solid coexistence, and solid state. In these three stages, the mechanical properties and thermophysical properties of the material change significantly, making the stress-strain constitutive relationship involved in the numerical analysis process particularly complex.
In the solidification process, three heat transfer modes are mainly involved: heat conduction, heat convection, and heat radiation. Heat conduction is the main heat transfer mode in the solidification process of the casting. Between the casting and the mold, heat is transferred through direct contact. The basic law of heat conduction satisfies Fourier’s law, and the heat conduction equation is:
$$\rho c_p \frac{\partial T}{\partial t} = \frac{\partial}{\partial x}\left( \lambda \frac{\partial T}{\partial x} \right) + \frac{\partial}{\partial y}\left( \lambda \frac{\partial T}{\partial y} \right) + \frac{\partial}{\partial z}\left( \lambda \frac{\partial T}{\partial z} \right) + Q \tag{23}$$
Heat convection is a heat transfer mode between the molten metal and the mold, and between the mold and the outside. The Newton cooling law describes heat convection:
$$q = a(T_f – T_w) \tag{24}$$
Heat radiation is also present during the solidification process of the casting, dissipating heat from the casting to the surrounding environment. The Stefan-Boltzmann law can be used to express the radiation heat transfer process:
$$q = \varepsilon \sigma_0 T_s^4 \tag{25}$$
The heat enthalpy method is an effective approach to handle the latent heat of solidification. In this method, the latent heat of solidification is treated as a temperature-dependent function. The latent heat quantity is determined by calculating the change in enthalpy during the period from the start of solidification to the completion of solidification, expressed as:
$$H = H_0 + \int_0^T C dT + (1 – f_s) L \tag{26}$$
The boundary condition settings in the simulation are crucial for accurate results. In actual production, the heat transfer capabilities between the casting and the mold, the casting and the cold iron, and the casting and the air are different. In the numerical simulation analysis of the solidification process, full consideration must be given to the setting of boundary conditions. Different heat transfer coefficients and interface grid conditions should be set between different entities. Typical interface heat transfer coefficients between different materials are listed in Table 2.
| Interface | Casting & mold | Casting & air | Mold & air | Casting & cold iron | Cold iron & mold |
|---|---|---|---|---|---|
| Coefficient | 300-1000 | 5-10 | 5-10 | 1000-5000 | 300-1000 |
For the interface mesh conditions in Visual-Mesh, two types exist: consistent meshes and non-consistent meshes. The consistent mesh (EQUIV) is used for the mesh treatment between different parts of the same entity, where both parts belong to the casting and have the same material properties, but were meshed separately, such as the casting body and the gating system, or the casting body and the riser. The two parts have the same medium and are continuous, with continuous temperature and velocity fields at the interface, and the nodes at the interface are shared by both meshes. If two parts have different materials but are welded together, EQUIV can also be used. COINC is used to define the heat exchange between two or more regions that do not intersect geometrically but are expected to be considered as connected in the simulation, mainly for interfaces between different materials with obvious differences in material properties and temperature, such as casting and mold, casting and cold iron, and casting and air. Therefore, to distinguish the temperature difference at the contact part of the two interfaces, the COINC boundary condition is selected to double the number of nodes at the interface. NCOINC is used to handle non-consistent (or non-matching) mesh interfaces. When adding different meshes together, the user must specify the interface as NCOINC. The NCOINC boundary condition allows the definition of heat transfer, fluid flow, etc., at the interface between two or more components, even if the meshes of these components do not match.
The common casting defects predicted by numerical simulation include gas porosity, cracks, inclusions, shrinkage porosity and shrinkage cavities, and segregation. Shrinkage porosity and shrinkage cavities are common defects in the casting process, mainly caused by the shrinkage of metal from the liquid state to the solidification process. Shrinkage porosity refers to dispersed and fine shrinkage holes formed in the last solidified region of the casting due to a lack of liquid metal or alloy feeding. Shrinkage cavities refer to concentrated, relatively large holes in the casting. The formation of shrinkage porosity and shrinkage cavities is related to factors such as the crystallization temperature range of the metal, the shape and size of the casting, pouring conditions, and mold design. The presence of shrinkage porosity and shrinkage cavities in castings reduces the mechanical properties of the casting and affects the service life and safety of the casting. The prediction criteria for shrinkage porosity and cavities include the Niyama criterion, the time shrinkage method, the critical solid phase method, and the temperature gradient method. The Niyama criterion is a common method for predicting shrinkage porosity and cavities. It is based on the relationship between the solid fraction and the temperature gradient during the solidification process of the casting. According to the Niyama criterion, when the solid fraction exceeds a critical value, if the temperature gradient is insufficient to maintain the feeding of the metal, shrinkage porosity and cavities will form. The commonly used form of the Niyama criterion is:
$$\frac{G}{\sqrt{R}} < C_{Niyama} \tag{27}$$
where G is the local temperature gradient and R is the solidification rate of the casting. The critical value CNiyama is generally taken as 1-1.1 for large castings and 0.8 for small castings.
Analysis and Optimization of Sand Casting of Compressor Support Ring Based on CAE Technology
The compressor support ring is a key component of the gas turbine compressor in thermal power plants. The compressor functions to compress the air and gas entering the gas turbine, thereby increasing the intake pressure and providing sufficient oxygen for the combustion chamber. Since it works in a high-temperature, high-pressure, high-humidity, and highly corrosive environment for a long time, the casting requirements for the compressor support ring are more stringent. Compared with ordinary castings, its size is larger, and the pouring time is longer. During the casting process, the molten metal is more likely to exhibit insufficient mold filling, or turbulent flow, resulting in inclusions, gas entrainment, shrinkage porosity, and shrinkage cavities. By simulating and analyzing the original process plan before actual pouring, possible quality problems in the casting can be predicted in advance, and the original plan can be optimized.
The casting studied is the lower half of a compressor support ring. The maximum external dimensions are 3070 mm × 1175 mm × 1462 mm, making it a medium-to-large casting with a weight of about 2.5 tons. The maximum wall thickness reaches 368 mm. The inner wall of the steam chamber is plated with a heat-resistant layer. The lower end of the casting body is integrally provided with a steam nozzle channel hole. Three feeding risers are arranged on the casting, and the side risers are provided with riser pads. This structure is characterized by reasonable design, simple structure, multiple functions, reliable connection of the inner cylinder, no noise, and good practicality.
The original casting process adopted by the enterprise is shown in Figure 2 in the original text. For medium-to-large castings, the bottom gating system is usually adopted. The mold placement is vertical. The pouring system consists of three ingates connected to the casting body. The pouring inlet is located on the lower side of the ring. The pouring cup size is 265 mm, and the cross-sectional diameter of the runner is 80 mm. Two waist-shaped cylindrical feeding risers with a size of 750 mm × 250 mm are arranged at the thick wall of the casting to prevent misrun at the lower bend of the compressor support ring. In addition, a circular feeding riser with a diameter of 450 mm is arranged at this position.
In the pre-processing stage, I completed the three-dimensional modeling of the casting body, feeding risers, pads, and gating system using NX11.0. The assembly was completed in NX11.0 and saved as an .igs file readable by Visual-Mesh. The assembly model was then imported into the Visual-Mesh application of ProCAST, and a box-shaped sand mold with dimensions of 4500 mm × 2500 mm × 2700 mm was added. After checking and repairing the surface gaps and cross-body problems of the model, the model orientation was adjusted to facilitate the subsequent addition of the gravity pouring direction. The model display format was set to wireframe and surface frame to check whether the merging of the casting and the mold sand box was problematic. For mesh generation, the sand box mesh size was set to 100, the runner and the connections between the runner and the casting were set to 20, and the casting part was set to 30. The surface mesh was checked and repaired, followed by the generation of the volume mesh. After automatic checking and repairing of the volume mesh, a total of 70,000 surface elements and 2.14 million volume elements were generated, with a total of approximately 2.23 million elements.
After mesh division, the gravity direction was set to the +Y direction. The casting material was ZG13Cr9Mo2Co1NiVNbNB (abbreviated as CB2), which is a new type of heat-resistant steel capable of serving under extreme conditions of 600-620 °C and 30 MPa. Since the ProCAST material database does not contain this alloy, I manually added the chemical composition and proportions of the material, and solved its physical performance parameters using the solver. The liquidus temperature was calculated as 1494 °C, and the solidus temperature as 1186 °C. The chemical composition of ZG13Cr9Mo2Co1NiVNbNB steel is shown in Table 3, and the room-temperature mechanical properties in Table 4.
| Item | C | Si | Mn | P | S | Cr | Ni |
|---|---|---|---|---|---|---|---|
| Specification | 0.11-0.14 | 0.20-0.30 | 0.80-1.00 | ≤0.020 | ≤0.010 | 9.00-9.60 | 0.10-0.20 |
| Yield strength Rm (MPa) | Tensile strength RP0.2 (MPa) | Elongation A (%) | Reduction of area Z (%) | Impact energy AkV2 (J) |
|---|---|---|---|---|
| ≥500 | 630-750 | ≥15 | ≥40 | ≥30a |
Note: a indicates the average value of three Charpy V-notch impact specimens, with a minimum value ≥ 24J.
The sand mold material was selected as silica sand. The thermal and physical parameters of the mold material, such as thermal conductivity and specific heat capacity, were taken from the material database. The casting, risers, sand mold, and gating system were then added to the volume manager. The casting body and gating system were defined as alloy parts with the material ZG13Cr9Mo2Co1NiVNbNB, the initial filling temperature was set to 1575 °C, and the initial fill fraction was set to 0. The stress type was chosen as linear elastic. The sand mold was defined as a mold part with the material Silica Sand-Permeable, the initial filling fraction was 100%, the initial temperature was 20 °C, and the stress type was rigid.
For the boundary conditions, the interface mesh between the casting body and the mold, and between the gating system and the mold, was set to COINC. The interface between the casting and the gating system was set to EQUIV because they share the same material and are continuous. The interface heat transfer coefficient between the ZG13Cr9Mo2Co1NiVNbNB alloy casting, the gating system, and the mold was set to 500 W/(m²·K). The heat boundary condition was set to air cooling on the outer surfaces of the mold, the riser, and the top surface of the pouring cup that contact air. The inlet condition was defined at the pouring cup with a mass flow rate of 100 kg/s, a filling temperature of 1575 °C, and a fill limit of 100%. The filling time was calculated as:
$$M_{ass} = V_{olume} \times \left( \frac{F_{ill} L_{imit}(\%)}{100} \right) \times D_{ensity}(F(T)) \tag{28}$$
$$F_{ill} T_{ime} = \frac{M_{ass}}{M_{ass} F_{low} R_{ate}} \tag{29}$$
The calculated filling time was approximately 107 s.
For the simulation parameters, the predefined parameters were selected as Gravity Filling. In the general simulation parameter settings, the simulation termination temperature was set to 1176 °C. To avoid the problem that the simulation ends before the number of steps is completed, the number of simulation steps was set as large as possible, i.e., 500,000 steps. The TFINAL and TENDFILL parameters were set to 0 to ensure that the termination step of the simulation process could be executed smoothly. The temperature simulation parameter settings enabled the thermal analysis module switch, and the TFREQ parameter was set to 50 to save a temperature field file every 50 steps. The flow field and stress field parameters were set according to the gravity casting template. After auto-checking the parameter settings, the simulation calculation was executed.
After the simulation was completed, the results were analyzed in Visual-Viewer. During the mold filling process, the temperature field of the molten metal was observed. At t = 0 s, the molten metal entered the sprue from the pouring gate, and the mold filling process began. At t = 4 s, the molten metal filled the entire gating system and began to enter the mold cavity. Due to the bottom gating system, the molten metal flowed relatively smoothly, with almost no significant scouring of the casting. When the filling proceeded to 48 s, the filling amount of the casting reached about 45%. With the continuous injection of high-temperature molten metal into the mold, the heat was enhanced, and the temperature remained basically stable. It could be observed that the temperature at the center position was higher than the temperature at the position where the surface contacted the sand mold. At 107 s, the mold filling was complete, and the filling amount of the casting reached 100%. It was relatively obvious that at the end of the mold filling process, the temperature field presented a bottom-to-top temperature gradient distribution, which was favorable for the directional solidification of the casting from bottom to top and the completion of riser feeding, thereby reducing the occurrence of defects such as shrinkage porosity and shrinkage cavities in the casting. The overall temperature range of the casting was 1575 °C-1510 °C, and the entire mold filling process was smooth without any misrun phenomenon.
In addition to the temperature field, the velocity field during the mold filling process was also analyzed. At 4.5% filling, the molten metal filled the entire gating system. The flow velocity of the molten metal in the sprue and cross runner was higher, about 2.5 m/s. The flow velocity of the middle ingate was higher than the other two, about 1.3 m/s, while that of the side ingates was about 0.5 m/s. In the subsequent 30%-90% filling process, the flow velocity in the mold cavity was stable at about 0.5 m/s, and the liquid level rose smoothly without obvious turbulence or gas entrainment. The velocity change curves at the center points of the three ingates showed that in the first 30 s of filling, the molten metal in the ingates was in a turbulent state with large fluctuations in flow velocity. After 30 s, the flow velocity gradually stabilized. The middle ingate showed particularly obvious changes. In the short term, the flow velocity increased rapidly to 2.4 m/s, then decreased to 0.9 m/s, and then rapidly increased to 1.5 m/s and became stable. The reason for this is that at the beginning of pouring, the molten metal could quickly fill the runner. Due to the large temperature difference between the inner wall of the runner and the molten metal, the molten metal adhered to the inner wall of the runner and solidified, resulting in a decrease in the cross-sectional area of the ingate and a decrease in the flow velocity. Along with the continuous entry of high-temperature molten metal into the ingate, the solid metal on the inner wall of the runner was washed away, allowing the flow velocity to return to a normal level.
During the solidification process, the solid fraction simulation results showed that since the solidification of the molten metal often occurs synchronously with the mold filling process, when the mold filling process had not yet ended, the edge and thin-wall parts of the casting had already begun to solidify. The solid fraction simulation results verified that when the filling reached 100%, the casting had a solid fraction of 6.9%. When the solid fraction reached 20%, the runner basically completed solidification, and the edges of the casting and risers gradually began to solidify. When the solid fraction exceeded 50%, the solidification proceeded from the outside to the inside, and the arc top center and the lower sides of the risers were the last to solidify. From the results, it could be concluded that the last solidified region was not entirely inside the riser, indicating that the riser did not achieve the expected feeding effect. The entire solidification process lasted for 37117 s.
From the temperature field simulation results of the solidification process, it could be observed that the temperature at the lower ends of the side risers connected to the pads and at the base platform under the arc top riser was higher than at other positions of the casting, and an ideal temperature gradient distribution had not been established. The top of the riser solidified earlier than the bottom, which would lead to insufficient feeding of molten metal after solidification, resulting in shrinkage porosity and shrinkage cavities. From the temperature-time curves and solid fraction-time curves at five different positions of the casting, the nodes at thick-wall positions cooled at a lower rate than the thin-wall positions. The solidification start time of the thin-wall positions was much earlier. Since the riser did not achieve directional and uniform solidification, the feeding channel of the riser was closed, and it could not provide continuous feeding for the solidification process at both ends of the casting. This situation would most likely cause insufficient feeding at the thick-wall positions, greatly increasing the possibility of shrinkage porosity and shrinkage cavities.
After the solidification was completed, the Niyama criterion was used to predict the casting defects. The defect prediction results of the original scheme showed that three large shrinkage cavities appeared at the lower ends of the three risers and extended into the interior of the casting. The reason was that the risers failed to achieve directional solidification from bottom to top, causing the feeding channel at the top to close, and the upper surface of the casting was not effectively fed. Large shrinkage porosity defects appeared on both sides of the middle riser, possibly due to the high temperature and high flow velocity of the molten metal at this position, leading to gas entrainment and oxidation. The volume of shrinkage porosity and shrinkage cavities accounted for about 16% of the total volume of the casting.
To address the defects predicted in the original scheme, I optimized the original casting process plan. First, in response to the three shrinkage cavity positions, in order to give full play to the feeding effect of the riser and establish a reasonable temperature gradient during solidification, the ordinary sand risers in the original plan were replaced with exothermic insulating risers. The operation method was to add a layer of FT400 exothermic insulating material with a thickness of 50 mm on the outside of the original riser. The riser sleeve thickness of the middle riser was 25 mm. The exothermic insulating riser can effectively avoid the appearance of shrinkage porosity and shrinkage cavities and improve the compactness of the casting. Its feeding efficiency can reach 30%-60%, far higher than that of ordinary sand risers. In response to the shrinkage porosity defects on both sides of the middle base platform, since the high-temperature molten metal entering the lower part of the mold cavity at the initial stage has a relatively high flow velocity, causing gas entrainment and oxidation, it is necessary to appropriately reduce the temperature of the molten metal at this position and increase the viscosity of the molten metal to reduce the flow velocity. Two pairs of external cold irons were placed on both sides of the base platform to accelerate the cooling rate of this part by absorbing the heat of the molten metal, thereby shortening the cooling time difference with other parts. The cold iron sizes were 740 mm × 210 mm × 130 mm and 217 mm × 200 mm × 156 mm, respectively. The improved casting process model was drawn using NX11.0 software, and the positions of the insulating sleeve and cold irons are schematically shown in the original text.
After optimizing the process, the model was re-imported into ProCAST for simulation. The surface mesh of the cold irons and the insulating sleeve was refined with a mesh size of 10. The outer surface of the sand mold was coarsened to 150 to reduce simulation time. A total of 53,033 surface elements and 1,162,161 volume elements were generated. In the pre-processing stage, the volume manager was updated to include the insulating sleeve and cold irons. The cold iron material was selected as low-carbon steel with an initial temperature of 20 °C. The interface heat transfer coefficient between the cold iron and the casting was set to 2000, and that between the insulating sleeve and the sand mold was set to 50. Other heat transfer coefficients remained the same as the original scheme.
The mold filling process of the new scheme was slightly longer, approximately 117 s. The simulation results showed that the molten metal flowed smoothly in the ingates, and the liquid level in the mold cavity rose stably. Under the action of the cold irons, the flow velocity of the molten metal on both sides of the middle riser was lower, and no liquid surface surging or gas entrainment occurred. At the end of the mold filling process, the temperature distribution of the molten metal in the entire mold cavity was relatively uniform, and no excessive temperature difference between the molten metal entering sequentially was observed. The mold filling integrity was excellent without cold shuts or misruns.
In the solidification process analysis of the new scheme, I added a zy cross-section along the X-axis of the casting to observe whether the three exothermic insulating risers could achieve the preset feeding effect. The simulation results showed that at the moment when the mold filling just ended, the solid fraction of the casting had already reached 12%. The edge, thin-wall parts, and the cold iron area of the casting began to solidify in the later stage of filling. When the solid fraction reached 60%, it was obvious that the casting solidified progressively from the thin-wall parts to the thick-wall parts. When the solid fraction increased from 80% to 95%, the molten metal solidified sequentially from the bottom to the top of the casting, and the final solidification positions were all located inside the exothermic insulating risers, ensuring that the risers provided effective feeding for the casting. The solidification process took about 24684 s. The temperature field distribution during solidification also showed that the cold irons had an effect. The thin-wall positions where the cold irons were placed had significantly lower temperatures than the thick-wall positions on both sides, forming an effective temperature gradient distribution. The temperature profile of the riser showed a bottom-to-top temperature distribution, and the molten metal achieved directional solidification, thereby effectively avoiding internal defects in the casting.
Using the Niyama criterion, the shrinkage porosity and shrinkage cavity prediction results of the optimized scheme were obtained. Compared with the original scheme, the shrinkage porosity and shrinkage cavities after solidification were mainly concentrated inside the risers, with a sporadic distribution in the runner system. No shrinkage defects occurred in the critical parts of the casting, proving that the optimized process plan effectively improved the shrinkage porosity and shrinkage cavities of the casting and improved the casting quality. However, a few small shrinkage porosity regions still existed at the bottom edge of the casting and at the connection with the middle ingate. Therefore, it was still necessary to further optimize the pouring process parameters to obtain the best casting process parameters.
Based on the new scheme, I designed orthogonal experiments to further optimize the casting pouring process parameters. The orthogonal experimental design is a design method that studies multiple factors and levels based on an orthogonal table to select representative test points for experimentation. These points have the characteristics of “uniform dispersion, neat comparability,” which can analyze primary and secondary factors, obtain the best combination and production process conditions, and guide the direction of the next experiment. Through the orthogonal experimental design, the influence of various process parameters on the casting quality can be analyzed in a relatively small number of experiments, the factors with the greatest influence can be identified, and the best casting process parameters can be selected. This not only reduces the test time and cost but also helps quickly find the multi-factor optimization scheme and effectively determine the best test parameters.
In this study, the three factors selected were pouring temperature, pouring speed, and sand mold temperature, each with three levels. The pouring temperature levels ranged from 1565 °C to 1585 °C, the pouring speed levels ranged from 90 to 105 kg/s, and the sand mold temperature levels were 20 °C, 25 °C, and 30 °C. The factor-level table is shown in Table 5. The L9(3⁴) orthogonal table was selected for the experiment. The experimental schemes are listed in Table 6.
| Level | (A) Pouring temperature / °C | (B) Pouring speed / (kg/s) | (C) Sand mold temperature / °C |
|---|---|---|---|
| 1 | 1585 | 100 | 20 |
| 2 | 1575 | 90 | 25 |
| 3 | 1565 | 105 | 30 |
| Test No. | A | Empty column | B | C | Experimental scheme |
|---|---|---|---|---|---|
| L1 | 1585 | 1 | 100 | 20 | A1B1C1 |
| L2 | 1585 | 2 | 90 | 25 | A1B2C2 |
| L3 | 1585 | 3 | 105 | 30 | A1B3C3 |
| L4 | 1575 | 1 | 90 | 30 | A2B2C3 |
| L5 | 1575 | 2 | 105 | 20 | A2B3C1 |
| L6 | 1575 | 3 | 100 | 25 | A2B1C2 |
| L7 | 1565 | 1 | 105 | 25 | A3B3C2 |
| L8 | 1565 | 2 | 100 | 30 | A3B1C3 |
| L9 | 1565 | 3 | 90 | 20 | A3B2C1 |
Using ProCAST simulation software, I sequentially performed simulation calculations for the nine test schemes in Table 6. After the simulation, the shrinkage porosity and shrinkage cavity prediction of the castings was viewed, and the shrinkage porosity rate of each group of tests was calculated using the defect volume. The calculated results were used as the only indicator to determine the best process parameters. The orthogonal experiment results are shown in Table 7.
| Test No. | A | B | C | Porosity / % |
|---|---|---|---|---|
| L1 | 1 | 1 | 1 | 15.34 |
| L2 | 1 | 2 | 2 | 14.36 |
| L3 | 1 | 3 | 3 | 16.63 |
| L4 | 2 | 1 | 2 | 13.37 |
| L5 | 2 | 2 | 3 | 10.53 |
| L6 | 2 | 3 | 1 | 12.97 |
| L7 | 3 | 1 | 3 | 17.63 |
| L8 | 3 | 2 | 1 | 14.67 |
| L9 | 3 | 3 | 2 | 18.52 |
For the analysis of multi-factor and multi-level experiments, range analysis is needed to evaluate the influence degree of different factors on the experimental results and find the optimal experimental condition combination. The core idea of range analysis is to calculate the range R of each factor at different levels (the difference between the maximum and minimum values). The larger the R value, the greater the influence of the factor on the experimental results, indicating that the factor is more important. The porosity range analysis results are shown in Table 8.
| Range | A | B | C |
|---|---|---|---|
| K1 | 46.33 | 46.34 | 42.98 |
| K2 | 36.86 | 39.56 | 46.25 |
| K3 | 50.82 | 48.12 | 44.79 |
| k1 | 15.44 | 15.45 | 14.33 |
| k2 | 12.29 | 13.19 | 15.42 |
| k3 | 16.94 | 16.04 | 14.93 |
| R | 13.96 | 8.56 | 3.27 |
From the range results, I found that R_A > R_B > R_C, indicating that the influence factors on the shrinkage porosity and shrinkage cavities of the compressor support ring casting, in descending order, are pouring temperature, pouring speed, and sand mold temperature. For the determination of the optimal scheme, since the goal of this experiment was to obtain the minimum porosity, the level corresponding to the smallest porosity index should be selected. Therefore, the optimal casting process parameter combination was A2B1C1, i.e., pouring temperature 1575 °C, pouring speed 100 kg/s, and sand mold temperature 20 °C.
Using the optimal pouring process parameters obtained from the orthogonal experiment, I performed a simulation analysis of the mold filling and solidification processes. The final prediction results of shrinkage porosity and shrinkage cavity positions after optimizing the process parameters are shown in the original text. It was intuitive that the shrinkage porosity and shrinkage cavity defect volume obtained by simulating with the new pouring process parameters was significantly smaller than that of the original scheme. The final porosity was about 8.4%, and the shrinkage defects after solidification were all located in the upper part of the risers, without affecting the quality of the casting itself. In addition, the small-area shrinkage porosity at the base platform and the middle ingate connection area in the original scheme was significantly improved. Overall, the casting effect of the large compressor support ring under this process parameter was excellent.
To verify the feasibility and effectiveness of the optimized casting process plan and the pouring parameters obtained from the orthogonal experiment, I carried out actual production of the compressor support ring. The pattern and sand mold were made according to the casting process drawing, and the structure and dimensions of the pattern and sand mold were checked. The melting of the molten metal was carried out using a medium-frequency furnace and a VOD refining furnace. The chemical composition of the molten metal was analyzed using a direct-reading spectrometer before pouring, and the results were within the composition deviation range of ZG13Cr9Mo2Co1NiVNbNB steel. Then, pouring was strictly carried out according to the parameters of pouring temperature 1575 °C, pouring speed 100 kg/s, and sand mold temperature 20 °C. After pouring, the casting was slowly cooled in the sand mold. After the cooling process was completed, the primary cleaning was performed, and the gates and risers were removed.
After cleaning, the performance heat treatment of the test blocks was carried out according to the procurement specification for ZG13Cr9Mo2Co1NiVNbNB steel castings for turbine cylinders and valve shells. Normalizing was carried out at 1110 °C ± 10 °C with a soaking time of 2 h + 26.5 h, and tempering at 740 °C ± 10 °C with a soaking time of 12 h + 28 h. After performance heat treatment, metallographic analysis was carried out at 200x and 400x magnification. The microstructure was tempered martensite, with grain sizes of 3-5 grades and 3.5-5 grades, respectively, which met the requirements of the specification.
After metallographic analysis, the second cleaning was carried out to remove excessive gates, residual risers, pads, etc., and to remove internal sintering and sand adhesion. Then, the casting was rough-machined and inspected by non-destructive testing. The non-destructive testing included VT (visual inspection), continuous MT (magnetic particle inspection), UT (ultrasonic inspection), and PT (penetrant inspection). The non-destructive testing results showed that no macroscopic defects were found in the casting, which was basically consistent with the simulation results above and met the requirements of the procurement specification. This demonstrates that the optimization of the compressor support ring process plan and pouring parameters was effective.
In summary, the main conclusions of this research are as follows:
1) Combined with the theory of sand casting process and the structural dimensions of the compressor support ring casting, the gravity sand casting method was selected for the compressor support ring. According to the characteristics of various pouring methods and the size of the casting, a bottom gating system was selected, and the positions and dimensions of the pads and risers were determined. The pouring process design of the steam chamber was completed.
2) The three-dimensional modeling of the mold was completed using modeling software. The finite element mesh and the pre-processing casting process parameters were set in ProCAST software, and the mold filling and solidification processes of the original scheme were simulated. Through analysis, it was found that in the early stage of filling with the original scheme, the initial flow velocity of the molten metal was too fast, causing a certain degree of gas entrainment and oxidation in the lower part of the casting. The solidification process did not form an effective temperature gradient distribution, resulting in three shrinkage cavities and two shrinkage porosity defects inside the casting. In response to the problems in the casting simulation of the original process, to control the flow velocity of the molten metal in the lower part and ensure a reasonable temperature gradient distribution so that the casting can achieve directional solidification, the optimization plan of replacing the ordinary risers with insulating risers and adding two groups of cold irons was determined.
3) Through the simulation analysis of the optimized new scheme, it was found that the cold irons provided a chilling effect, and the shrinkage porosity defects on both sides of the middle riser of the casting were significantly reduced. The exothermic insulating risers helped the riser area achieve directional solidification from bottom to top, and the casting was effectively fed. The shrinkage cavities that appeared in the interior of the casting in the original scheme were all located inside the exothermic insulating risers. To further eliminate the small shrinkage porosity at the connection between the middle ingate and the casting and the base platform area, orthogonal experiments were designed based on the optimized scheme to further optimize the casting pouring process parameters.
4) The three pouring parameters that greatly influence the casting quality, namely pouring temperature, pouring speed, and sand mold temperature, were used as three factors to design a three-factor, three-level simulation orthogonal experiment. Nine combination experiments were conducted, and the shrinkage porosity rate obtained from the simulation was used as the only indicator for determining the best process parameters. Through direct analysis of the orthogonal experiment design results, the influence factors on the shrinkage porosity rate of the compressor support ring casting, in descending order, were pouring temperature, pouring speed, and sand mold temperature. The optimal pouring process parameter combination was pouring temperature 1575 °C, pouring speed 100 kg/s, and sand mold temperature 20 °C. The simulation analysis using this process parameter combination showed excellent casting quality with no shrinkage porosity or shrinkage cavities. The simulation results were then verified through actual production. Non-destructive testing technology was used to inspect the casting, and no macroscopic defects were found in the casting, proving the effectiveness of the optimization of the compressor support ring process plan and pouring parameters.
Conclusion and Outlook
This thesis combined the actual casting problems faced by foundry enterprises in the production process and used the casting simulation software ProCAST to conduct a simulation study on the casting process of a certain type of compressor support ring. Based on the simulation analysis of the mold filling and solidification processes, the temperature field, solid fraction, and defect prediction, the casting process plan was redesigned and optimized, and the main pouring process parameters were adjusted through orthogonal experiments. The above series of work provided reference for the actual casting production. However, the use of casting simulation software cannot fully reproduce actual production conditions, and the simulation process still has many aspects that need to be improved. Therefore, further research on the compressor support ring steel casting process is still needed.
First, the material database of ProCAST needs to be improved. To withstand the high-temperature and high-pressure environment, steam chamber castings for thermal power and nuclear power are usually made of heat-resistant steel grades such as CB2, T91, P91, and H53. These grades are not included in the ProCAST database. The database also lacks some mold sand and core sand materials. Therefore, a complete and comprehensive material database needs to be established in the future.
Second, the simulation parameters need to be adjusted. The composition of the ZG13Cr9Mo2Co1NiVNbNB heat-resistant steel used in the casting was manually added, and its physical parameters such as solidus temperature, liquidus temperature, thermal conductivity, and shrinkage rate were automatically generated by the ProCAST software, which may differ from the actual physical properties of the material. If the simulated parameters are used in actual production, the results may differ from the simulation results. Therefore, it is recommended to verify the material performance parameters through experimental means to reduce the error between the simulation results and actual production.
Third, the microscopic structure simulation analysis should be increased. In this paper, only macroscopic simulations of the compressor support ring were carried out, such as temperature field, velocity field, solid fraction, and shrinkage porosity. However, the microscopic structure of the casting was not simulated and analyzed. With the help of the CAFE algorithm in ProCAST software, the influence of the changes of various parameter values on the solidification microstructure of ZG13Cr9Mo2Co1NiVNbNB heat-resistant steel castings can be analyzed.
Finally, as power generation equipment progresses from ultra-supercritical to ultra-ultra-supercritical units, it is also necessary to further study the casting process of castings for new power generation units, providing support for the sustainable development of the enterprise and the progress of casting technology.