A method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots

Abstract

This invention belongs to the field of metal smelting technology and relates to a method for predicting alloy element segregation in large-scale titanium alloy VAR ingots. The method includes: acquiring VAR furnace structure data; establishing a macroscopic FE model based on the VAR furnace structure data to calculate the ingot temperature field distribution; acquiring microscopic CA grids and state parameters; establishing a microscopic CA model using the CA grids and state parameters to calculate the microscopic solute concentration distribution; mapping the ingot temperature field distribution to the microscopic CA model for FE-CA cross-scale coupling to generate an FE-CA cross-scale coupled model for simulating the VAR process under different process parameter combinations, and generating an optimal process parameter combination with the objective that the maximum segregation degree of the target element should not be less than a preset threshold. This invention simulates the solidification process under different process conditions to predict the occurrence and development of compositional segregation in large-scale VAR titanium alloy ingots, thereby reducing segregation.

Description

Technical Field

[0001] This invention relates to the field of metal smelting technology, and in particular to a method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots. Background Technology

[0002] Titanium and titanium alloys possess high specific strength, corrosion and heat resistance, non-magnetic and non-toxic properties, and good biocompatibility. They are important metallic materials after steel and aluminum, widely used in aerospace, marine, military, nuclear power, electronics, medical, and chemical industries, and are known as the "third metal," "space metal," and "marine metal." Due to the high chemical reactivity of titanium alloys, their smelting is generally carried out under vacuum or inert atmosphere conditions. Vacuum arc remelting technology is simple to operate, with a relatively fast melting rate of alloying elements and minimal volatilization loss. It can also produce large ingots that meet industrial needs. After more than 60 years of development, vacuum arc remelting technology has reached maturity and is currently one of the main methods for smelting titanium and titanium alloy ingots. During the smelting process, the melting current, melting rate, and cooling conditions are not constant. Therefore, improper control of process parameters can easily lead to problems such as alloy element segregation, uneven microstructure, and porosity and shrinkage cavities, significantly affecting subsequent processing and the performance of the finished product. Due to the high cost of industrial testing in titanium alloy production, and the difficulty in observing the physicochemical phenomena and ingot microstructure during vacuum consumable melting, numerical simulation methods have gradually replaced traditional trial-and-error experiments and become the mainstream means of studying the solidification behavior of titanium alloys. This has helped to optimize ingot production processes, improve ingot quality, and provide some guidance.

[0003] A prior art method discloses a numerical simulation method for titanium alloy segregation during electron beam cold hearth melting, belonging to the field of metal solidification process simulation. The method includes: constructing a three-dimensional model of a crystallizer; obtaining the performance parameters of the molten titanium alloy and the three-dimensional model of the crystallizer using JmatPro material property simulation software, and inputting the heat transfer coefficient of the molten titanium alloy into the three-dimensional model of the crystallizer to obtain the crystallizer model; establishing a fluid flow field-solidification-melting-component transport model using ANSYS simulation software, and solving the crystallizer model using the model to obtain the numerical simulation results of titanium alloy segregation. This method effectively improves the computational efficiency of the segregation prediction mathematical model, and the simulation method can reduce the initial cost of production experiments. Compared with industrial test methods, laboratory test methods, and numerical simulation methods, it provides more accurate and faster results in theoretical analysis.

[0004] The existing technical solution also discloses an intelligent control method for the stirring intensity of VAR melting of titanium alloys. This technical solution relates to the field of metal smelting technology and includes the following steps: S1, calculating the distribution of stirring forces during the VAR melting process; S2, calculating the velocity field and temperature field of the molten pool based on the force distribution; S3, constructing an eddy current field distribution model based on the velocity field and temperature field; S4, optimizing the electromagnetic stirring force field of the molten pool based on the eddy current field; S5, adjusting the arc scanning mode based on the optimized electromagnetic stirring force field; S6, performing real-time control of the arc scanning based on dynamic flow feedback. By setting an electromagnetic stirring control model based on eddy current field optimization, precise control of the molten pool flow is achieved, improving stirring uniformity and optimizing melting quality. This technical solution uses unsteady-state electromagnetic stirring force calculation, combined with eddy current induction effect to dynamically adjust the stirring intensity, making the molten pool velocity distribution more uniform.

[0005] However, the existing technical solutions mentioned above lack a systematic, multi-field coupled numerical simulation process for the VAR melting process of titanium alloys, which makes it more difficult to observe and predict compositional segregation due to the complex production process. Summary of the Invention

[0006] To address the problems existing in the prior art, the present invention discloses a method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots. This method simulates the solidification process under different process conditions and parameters, predicts the occurrence and development of compositional segregation in large-size VAR titanium alloy ingots, and optimizes process parameters based on simulation results to reduce the generation of segregation.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] A method for predicting alloy element segregation in large-size titanium alloy VAR ingots includes:

[0009] Obtain VAR furnace structure data, and based on the VAR furnace structure data, establish a macroscopic FE model to calculate the temperature field distribution of the billet.

[0010] Obtain the microscopic CA grid and state parameters, and establish a microscopic CA model using the CA grid and state parameters to calculate the microscopic solute concentration distribution;

[0011] The temperature field distribution of the billet is mapped to the microscopic CA model to perform FE-CA cross-scale coupling, generating an FE-CA cross-scale coupling model for simulating the VAR process under different combinations of process parameters. The optimal combination of process parameters is generated with the goal that the maximum segregation of the target element should not be less than a preset threshold.

[0012] Optionally, establishing the macroscopic FE model includes:

[0013] Based on the VAR furnace structure data, a geometric model including electrodes, arc zone, ingot, and crystallizer is established, and an adaptive tetrahedral mesh is used to locally refine the solid-liquid interface region to generate the macroscopic FE model.

[0014] Optionally, calculating the temperature field distribution of the billet includes:

[0015] The arc heating power density of the macroscopic FE model is solved using the A-φ method, and the spatial distribution of arc heating power is obtained by combining the current and voltage.

[0016] The initial temperature and boundary conditions are set, and a transient temperature field model including arc heating, latent heat release and boundary heat exchange is established based on the spatial distribution of the electric arc heating power and the latent heat release law, which is used to calculate the temperature field distribution of the ingot.

[0017] Optionally, establishing the transient temperature field model includes:

[0018] ;

[0019] in, Density of titanium alloy For specific heat capacity, Thermal conductivity, For latent heat of solidification, Spatial distribution of electric arc heating power. Instantaneous temperature For smelting time, This is the vector gradient operator.

[0020] Optionally, the boundary conditions include: fixed heat flux boundary, convective heat dissipation boundary, convective heat dissipation boundary, and adiabatic boundary.

[0021] Optionally, the microscopic CA model includes: a grain growth sub-model and a multi-element solute diffusion sub-model;

[0022] The grain growth sub-model adopts a random nucleation mechanism based on supercooling, and the solid-liquid interface growth rate is proportional to the supercooling, which is used to describe the grain growth state.

[0023] The multi-element solute diffusion sub-model considers the difference in diffusion coefficients between the solid and liquid phases and the equilibrium of solute distribution, and is used to calculate the microscopic solute concentration distribution.

[0024] Optionally, the nucleation probability of the random nucleation mechanism includes:

[0025] ;

[0026] in, For the probability of nucleation, For nucleation density, For time steps, The volume of a cell. The critical nucleation free energy, Boltzmann's constant, The temperature is the local temperature at the nucleation moment.

[0027] Optionally, calculating the microscopic solute concentration distribution includes:

[0028] ;

[0029] in, Let i be the concentration of element i. Where is the diffusion coefficient. For diffusion time, This is the gradient operator.

[0030] Optionally, generating the FE-CA cross-scale coupling model includes:

[0031] The macroscopic FE model is used to solve the temperature field distribution of the billet according to the target time step, and the temperature history of each FE unit is extracted. The temperature history is mapped to the CA grid corresponding to the microscopic CA model through bilinear interpolation. This is used to update the phase state and grain growth state based on the temperature history and to calculate the microscopic solute concentration distribution.

[0032] The solute redistribution amount at the solid-liquid interface is determined using the microscopic solute concentration distribution, which is then fed back to the macroscopic FE model to correct the latent heat release and optimize the solution of the billet temperature field distribution.

[0033] The macroscopic FE model and the microscopic CA model use the same time step to ensure the spatiotemporal synchronization of the macroscopic physical field and the microscopic organizational evolution.

[0034] The beneficial effects of this invention are as follows:

[0035] 1. Achieving unified modeling of macroscopic process parameters and microscopic segregation behavior to improve the physical consistency of segregation prediction: This invention establishes a cross-scale coupling mechanism between a finite element (FE) model and a cellular automata (CA) model, using the temperature field as the information transmission medium. This enables a bidirectional correlation between macroscopic electromagnetic-thermal processes and microscopic grain growth and solute redistribution behavior during vacuum consumable melting. Compared to existing methods based solely on macroscopic heat transfer models or single microscopic models, this invention can simultaneously reflect the comprehensive impact of process parameter variations on molten pool morphology, solidification conditions, and alloy element segregation behavior within the same computational framework, thereby improving the physical consistency and reliability of segregation prediction results.

[0036] 2. Quantitative prediction of multi-element segregation degree and spatial distribution, avoiding reliance on empirical judgment: This invention introduces a multi-element solute diffusion and distribution mechanism into the CA model, enabling simultaneous simulation of the migration and redistribution behavior of multiple alloying elements such as Al, V, Sn, Zr, and Mo during solidification, and outputting the corresponding segregation coefficient, maximum segregation degree, and spatial distribution characteristics. Compared with prediction methods relying on empirical formulas or single-element approximations, this invention can quantitatively describe the differences in segregation behavior of different elements in large-size VAR ingots, which is beneficial for identifying high-risk segregation regions and provides a clear basis for subsequent process adjustments and quality assessments.

[0037] 3. It can reflect the influence of process parameter variations on segregation and guide process optimization: By inputting different combinations of melting current, voltage, casting speed, and cooling intensity parameters into the FE-CA coupled model, this invention can systematically analyze the influence trends of process parameter variations on solidification conditions and alloy element segregation behavior. This technical advantage allows the invention not only to predict segregation results but also for comparative analysis and optimization of process parameters, thereby reducing the trial-and-error costs of traditional methods relying on multiple industrial trials and improving the efficiency and relevance of process design.

[0038] 4. Applicable to large-diameter titanium alloy VAR ingots, with strong engineering applicability: The modeling method used in this invention can adapt to the melting conditions of large-diameter VAR ingots and can be parametrically adjusted according to the actual equipment size and process conditions, avoiding the problem of being unable to be used in industrial scenarios due to model size limitations. Therefore, this invention, while ensuring calculation accuracy, has good engineering scalability and is suitable for production process analysis and quality control of large-diameter titanium alloy VAR ingots.

[0039] 5. It helps reduce production risks and improve the stability of finished product performance: By predicting the location and extent of alloy element segregation in advance during the smelting stage, this invention can provide a reference for subsequent homogenization heat treatment and process route formulation, reduce the risk of performance fluctuations caused by segregation, and improve the microstructure uniformity and performance stability of large-size titanium alloy ingots. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of a method for predicting the segregation of alloying elements in a large-size titanium alloy VAR ingot according to an embodiment of the present invention.

[0042] Figure 2This is a mesh generation diagram of the FE model for a large-size titanium alloy VAR ingot according to an embodiment of the present invention;

[0043] Figure 3 This is a temperature field cloud diagram of the Ti-6Al-4V alloy according to an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram of the phase distribution and grain growth of the CA model in an embodiment of the present invention;

[0045] Figure 5 This is a spatial distribution diagram of Al elements according to an embodiment of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1 As shown in the figure, this embodiment discloses a method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots, including: acquiring VAR furnace structure data; establishing a macroscopic FE model based on the VAR furnace structure data to calculate the temperature field distribution of the ingot; acquiring microscopic CA grids and state parameters; establishing a microscopic CA model using the CA grids and state parameters to calculate the microscopic solute concentration distribution; mapping the ingot temperature field distribution to the microscopic CA model for FE-CA cross-scale coupling to generate an FE-CA cross-scale coupling model to simulate the VAR process under different combinations of process parameters, and generating the optimal combination of process parameters with the goal that the maximum segregation degree of the target element should not be less than a preset threshold.

[0049] Specifically, the core of this invention is to construct an FE-CA cross-scale coupling model, which realizes the bidirectional transmission of macroscopic process parameters and microstructure evolution through the temperature field, and specifically includes the following steps:

[0050] Step 1: Establish a macroscopic FE model (electromagnetic field + temperature field):

[0051] 1.1 Geometry and Mesh Modeling: Based on the actual structure of the VAR equipment, a geometric model including electrodes, arc zone, ingot, and crystallizer is established. Tetrahedral elements are used to divide the mesh (the number of meshes for large-size ingots is ≥500,000) to ensure the mesh density in areas with large temperature gradients (solid-liquid interface).

[0052] 1.2 Electromagnetic field governing equations: Based on Maxwell's equations, the A-φ method is used to solve for the electromagnetic field distribution in the arc region and calculate the arc heating power density.

[0053] ;

[0054] ;

[0055] in, Vector magnetic potential. For scalar potential, Permeability, The electric arc conductivity, For current density, For time, As a gradient operator, the spatial distribution of arc heating power is obtained by inputting current and voltage parameters. .

[0056] 1.3 Temperature Field Control Equations: Based on the spatial distribution of arc heating power (as a volumetric heat source) and the law of latent heat release (the law of heat conduction), considering arc heating, latent heat release, and boundary heat dissipation, the transient temperature field equations are established as follows:

[0057] ;

[0058] in, Density of titanium alloy For specific heat capacity, Thermal conductivity (varying with phase and temperature). The latent heat of solidification (determined by the equivalent specific heat capacity method or the volume fraction method). Corresponding to electric arc heating, Corresponding to latent heat release, boundary heat dissipation is controlled by boundary conditions. This refers to the instantaneous temperature.

[0059] 1.4 Boundary conditions:

[0060] Arc-ingot interface: fixed heat flux boundary, provided by electromagnetic field solution results;

[0061] Crystallizer-billet interface: Convection heat dissipation boundary, the formula for convection heat dissipation is:

[0062] ;

[0063] This refers to the temperature of the side wall of the ingot. Cooling medium temperature, 20~40℃, heat dissipation coefficient Calibrated by cooling intensity, ;

[0064] Bottom of the billet: convection heat dissipation boundary, the convection heat dissipation formula is:

[0065] ;

[0066] Temperature at the bottom of the billet, heat dissipation coefficient Calibrated by cooling intensity, ;

[0067] Central axis: adiabatic boundary (radial temperature gradient is 0).

[0068] Step 2: Establish a microscopic CA model (grain growth + solute distribution):

[0069] 2.1 CA Mesh and State Parameters: A three-dimensional cubic cellular mesh with cell sizes of 1~10μm was adopted. Each FE macroscopic unit (size 5~20mm) corresponds to N×N×N (N=200~500) CA cells, and a coordinate mapping relationship was established. Initial state parameters were assigned to each cell (initial phase is liquid, initial solute concentration is the alloy average concentration, and grain identification is 0 (no nucleation)). Through FE-CA coupling, the temperature history (T(x, y, z, t)) of the macroscopic FE model was imported into each CA cell as the core driving condition for phase update, grain growth, and solute diffusion. The state parameters of each cell were updated step by step until all cells were transformed into the solid phase, completing the microscopic evolution simulation.

[0070] CA cell state parameters include:

[0071] Phase: Liquid phase solid phase Solid-liquid coexistence state ,in Liquidus temperature This refers to the solidus temperature.

[0072] Grain identification: A unique number distinguishes different grains;

[0073] Solute concentration: The mass fraction of multiple elements (such as Al, V, Sn, etc.).

[0074] 2.2 Grain growth model, used to describe the constraint of grain morphology on solute diffusion, to simulate solute redistribution at the solid-liquid interface:

[0075] Nucleation Model: A stochastic nucleation mechanism based on undercooling is adopted, and the nucleation probability P satisfies:

[0076] ;

[0077] in, nucleation density (10) 5~10 6 m -3 ), For time steps, The volume of a cell. The critical nucleation free energy, is the Boltzmann constant.

[0078] Growth Model: Growth Rate at the Solid-Liquid Interface With supercooling Proportional:

[0079] ;

[0080] in, The growth rate coefficient (10) -6 m / (s·K)).

[0081] 2.3 Multi-element solute diffusion model:

[0082] Based on Fick's law, considering the difference in diffusion coefficients between the solid and liquid phases and the equilibrium of solute distribution, the solute diffusion equation is solved:

[0083] ;

[0084] in, Let i be the concentration of element i. Diffusion coefficient (liquid phase) solid phase ).

[0085] The solid-liquid interface satisfies the equilibrium distribution relationship:

[0086] ;

[0087] in, The solute partition coefficient of element i (e.g., Al: 0.95, V: 0.90). These are the equilibrium concentrations of element i in the solid and liquid phases, respectively.

[0088] Step 3: FE-CA cross-scale coupling mechanism:

[0089] Using the temperature field as the core coupling medium, a two-way data transfer between macroscopic and microscopic levels is achieved. The specific process is as follows:

[0090] Temperature field transfer: FE model by time step Solve for the temperature field and extract the temperature history of each FE element. The temperature input to the CA model is mapped to the corresponding CA grid through bilinear interpolation.

[0091] Microscopic state feedback: The CA model updates the phase state and grain growth state based on the temperature field, solves the solute diffusion equation to obtain the microscopic solute distribution, and calculates the solute redistribution at the solid-liquid interface. Feedback is fed back to the FE model to correct the latent heat release. Optimize the accuracy of temperature field solution;

[0092] Time step synchronization: The FE model and the CA model use the same time step to ensure the spatiotemporal synchronization of the macroscopic physical field and the microscopic organization evolution.

[0093] Step 4: Segregation Prediction and Process Optimization

[0094] 4.1 Segregation Quantification Indicators: Based on the microscopic solute concentration distribution output by the CA model, macroscopic segregation evaluation indicators are calculated:

[0095] Segregation coefficient: ,in Let i be the solute concentration in cell i. This represents the average solute concentration of the alloy.

[0096] Maximum segregation: ;

[0097] Segregation homogeneity: ;

[0098] in, For the first i The target element in the first j Segregation coefficients in a CA cell The total number of CA cells involved in the segregation uniformity calculation.

[0099] 4.2 Simulation of multiple process parameters:

[0100] Cyclic input of different process parameter combinations: equivalent melting current Ie (30~35, normalized parameter) and equivalent voltage Ue (30~35, normalized parameter); casting speed 8~10mm / min; cooling intensity 15~25m. 3 / h, through the coupling model, outputs segregation indices and establishes a database of process parameters-segregation relationships.

[0101] 4.3 Process Optimization: Based on the database, the optimal parameter combination is selected to maximize the segregation of the target element. This meets the performance requirements for large-size titanium alloy ingots. When multiple sets of parameters meet... When, prioritize The smallest possible combination ensures the overall uniformity of the ingot composition and reduces the risks associated with subsequent processing.

[0102] This embodiment discloses a method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots, including:

[0103] S1. Establish a macroscopic finite element (FE) model, which is used to solve the electromagnetic field and temperature field in the vacuum arc remelting (VAR) process. The input process parameters include cooling intensity, casting speed, current and voltage.

[0104] S2. Establish a microscale cellular automaton (CA) model, which is used to simulate the grain growth and multi-element solute distribution of titanium alloys. The input of the CA model includes the temperature field data output by the FE model.

[0105] S3. Construct an FE-CA cross-scale coupling mechanism to achieve bidirectional data transfer between macro and micro scales through the temperature field, and synchronously update the temperature field of the FE model and the grain growth and solute distribution state of the CA model.

[0106] S4. Based on the coupled FE-CA model, simulate the VAR process under different combinations of process parameters, and output the multi-element segregation prediction results of titanium alloy VAR ingots. The segregation prediction results include the segregation coefficient, the maximum segregation degree, and the spatial distribution of the target element.

[0107] Further, in step S1, the diameter of the large-size titanium alloy VAR ingot is ≥500mm, and the titanium alloy includes at least one of Ti-6Al-4V and Ti-5Al-2Sn-2Zr-4Mo-4Cr, and the multi-element includes at least two of Al, V, Sn, Zr, Mo and Cr.

[0108] Further, in step S1, the electromagnetic field control equations of the FE model are based on Maxwell's equations and solved using the A-φ method to obtain the arc heating power distribution; the temperature field control equations are based on the law of heat conduction, considering arc heating, latent heat release, and boundary heat dissipation, and are solved to obtain the temperature history at different locations of the ingot, including the liquidus temperature T. L and solidus temperature T S .

[0109] Further, in step S2, the cell size of the CA model is 1~10μm, and each FE macroscopic unit corresponds to Each CA cell maps the FE temperature field to the CA grid using bilinear coordinate interpolation.

[0110] Furthermore, in step S2, the grain growth model adopts a random nucleation mechanism based on supercooling and a linear growth rate model. The nucleation probability is positively correlated with the nucleation density and supercooling, and the growth rate is proportional to the supercooling. The multi-element solute distribution model is based on Fick's law and considers the solute partition coefficient k of each element. i Liquid phase diffusion coefficient D i,L and solid-phase diffusion coefficient D i,S .

[0111] Further, in step S3, the FE-CA cross-scale coupling mechanism includes:

[0112] S31.FE model by time step Solve the temperature field and extract the temperature data for each FE macroscopic element. ;

[0113] S32. The temperature data is mapped to the corresponding CA microgrid, and the CA model determines the cell phase (liquid phase, solid phase, solid-liquid coexistence state) based on the temperature.

[0114] The S33.CA model updates the grain growth state based on phase state, such as... Figure 4 As shown, the solute concentration distribution is obtained by solving the solute diffusion equation. Figure 5 As shown, calculate the amount of solute redistribution at the solid-liquid interface;

[0115] S34. Feed the solute redistribution amount back to the FE model to correct the latent heat release and update the temperature field;

[0116] S35. Repeat steps S31 to S34 to achieve closed-loop coupling between the FE model and the CA model.

[0117] Further, in step S4, the segregation coefficient Maximum segregation ,when When it is a positive bias analysis, The time is a negative bias.

[0118] Further, in step S5: based on the segregation prediction results, optimize the process parameters to maximize the segregation degree of the target element. .

[0119] In one embodiment, a Ti-6Al-4V titanium alloy VAR ingot with a diameter of Φ800mm was used as the object, and the segregation prediction of Al and V elements was performed using the method described in this invention.

[0120] Step 1: Establish a macroscopic FE model:

[0121] Geometric model: Based on the actual VAR furnace structure, an axisymmetric two-dimensional geometric model is established, including electrodes, arc zone, ingot, and crystallizer, with an axial height of 2000mm.

[0122] Mesh generation: An adaptive tetrahedral mesh was used, with a total mesh count of approximately 550,000. Local refinement was applied in the solid-liquid interface region, such as... Figure 2 As shown.

[0123] Electromagnetic field solution: Input current I = 30kA, voltage U = 40V, the power density distribution of electric arc heating is obtained by using the A-φ method.

[0124] Temperature field solution: Based on the established transient temperature field equations, initial temperature (room temperature) and boundary conditions (fixed heat flow boundary, convective heat dissipation boundary, adiabatic boundary) are set, and the temperature field distribution of the billet is obtained through numerical solution, such as... Figure 3 As shown.

[0125] Step 2: Establish a micro-CA model:

[0126] The CA mesh size is set to 5 μm, with each FE macroscopic unit corresponding to 300×300×300 CA cells. Nucleation density. growth rate coefficient The partition coefficients for Al and V are k=0.95 and k=0.90, respectively. The liquid phase diffusion coefficients are taken as follows: and .

[0127] Step 3: FE-CA cross-scale coupling:

[0128] Time step The FE temperature field is interpolated to the CA grid at each step. The CA model updates the phase state and solute distribution, and feeds back the latent heat correction to the FE model, achieving two-way coupling.

[0129] Step 4: Segregation Prediction and Optimization

[0130] Simulation results show that the maximum segregation coefficient K of Al element max =1.18, V element K max =1.22, indicating significant segregation.

[0131] By adjusting process parameters (current reduced to 28kA, cooling intensity increased to...), After resimulation, Al element K max The value of V decreases to 1.12, and the value of V decreases to 1.14, satisfying K. max The requirement is ≤1.15.

[0132] In one embodiment, a Ti-5Al-2Sn-2Zr-4Mo-4Cr (Ti-17) alloy ingot with a diameter of 600 mm was used as the object, and the process parameters were optimized through simulation.

[0133] Step 1: Set multiple sets of process parameters, as shown in Table 1.

[0134] Table 1

[0135]

[0136] Step 2: Perform FE-CA coupled simulations separately: run a complete solidification process simulation once for each set of parameters; record the segregation indices of the four elements Al, Sn, Zr, and Mo.

[0137] Step 3: Result comparison and optimization selection, as shown in Table 2.

[0138] Table 2

[0139]

[0140] Step 4: Optimization conclusion: The second set of parameters is selected as the recommended process because its maximum segregation is ≤1.18 and its uniformity is the best. After adopting this set of parameters in actual production, the uniformity of the ingot composition is improved and the subsequent heat treatment time is shortened by about 15%.

[0141] In summary, this implementation offers the following advantages: 1. High prediction accuracy: The FE-CA cross-scale coupling model considers both macroscopic process parameter influences and microscopic solute redistribution mechanisms, achieving a prediction error of ≤5% compared to experiments, significantly outperforming traditional single-scale models; 2. Multi-scenario adaptability: Applicable to large-size titanium alloy VAR ingots with diameters of 500~1000mm, supporting various multi-element titanium alloy systems such as Ti-6Al-4V, and adaptable to different VAR equipment parameters; 3. Strong guidance for process optimization: Quantitatively reveals the synergistic effects of current, voltage, casting speed, and cooling intensity on multi-element segregation, providing direct evidence for process parameter optimization and reducing experimental trial-and-error costs by ≥30%; 4. Significant engineering value: Can predict segregation risk areas in advance, guiding subsequent homogenization heat treatment process design and improving the yield and performance stability of large-size titanium alloy ingots; 5. Strong scalability: The model can be extended to other melting and casting processes (such as ESR, VAR+VAR dual melting) and other alloy systems (such as high-temperature alloys and stainless steel), with broad application prospects.

[0142] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for predicting the segregation of alloying elements in large-size titanium alloy VAR ingots, characterized in that, include: Obtain VAR furnace structure data, and based on the VAR furnace structure data, establish a macroscopic FE model to calculate the temperature field distribution of the billet. The calculation of the temperature field distribution of the billet includes: The arc heating power density of the macroscopic FE model is solved using the A-φ method, and the spatial distribution of arc heating power is obtained by combining the current and voltage. The initial temperature and boundary conditions are set, and a transient temperature field model including electric arc heating, latent heat release and boundary heat exchange is established based on the spatial distribution of electric arc heating power and the law of latent heat release, which is used to calculate the temperature field distribution of the billet. Obtain the microscopic CA grid and state parameters, and establish a microscopic CA model using the CA grid and state parameters to calculate the microscopic solute concentration distribution; The microscopic CA model includes: a grain growth sub-model and a multi-element solute diffusion sub-model; The grain growth sub-model adopts a random nucleation mechanism based on supercooling, and the solid-liquid interface growth rate is proportional to the supercooling, which is used to describe the grain growth state. The multi-element solute diffusion sub-model considers the difference in diffusion coefficients between the solid and liquid phases and the equilibrium of solute distribution, and is used to calculate the microscopic solute concentration distribution. The temperature field distribution of the billet is mapped to the microscopic CA model to perform FE-CA cross-scale coupling, generating an FE-CA cross-scale coupling model for simulating the VAR process under different combinations of process parameters. The optimal combination of process parameters is generated with the goal that the maximum segregation of the target element should not be less than a preset threshold.

2. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, Establishing the macroscopic FE model includes: Based on the VAR furnace structure data, a geometric model including electrodes, arc zone, ingot, and crystallizer is established, and an adaptive tetrahedral mesh is used to locally refine the solid-liquid interface region to generate the macroscopic FE model.

3. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, Establishing the transient temperature field model includes: ; in, For the density of titanium alloy, For specific heat capacity, Thermal conductivity, For latent heat of solidification, Spatial distribution of electric arc heating power. Instantaneous temperature For smelting time, This is the vector gradient operator.

4. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, The boundary conditions include: fixed heat flux boundary, convective heat dissipation boundary, convective heat dissipation boundary, and adiabatic boundary.

5. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, The nucleation probability of the random nucleation mechanism includes: ; in, For the probability of nucleation, For nucleation density, For time steps, The volume of a cell. The critical nucleation free energy, Boltzmann's constant, The temperature is the local temperature at the nucleation moment.

6. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, Calculating the microscopic solute concentration distribution includes: ; in, Let i be the concentration of element i. The diffusion coefficient is... For diffusion time, This is the gradient operator.

7. The method for predicting alloy element segregation in large-size titanium alloy VAR ingots according to claim 1, characterized in that, Generating the FE-CA cross-scale coupling model includes: The macroscopic FE model is used to solve the temperature field distribution of the billet according to the target time step, and the temperature history of each FE unit is extracted. The temperature history is mapped to the CA grid corresponding to the microscopic CA model through bilinear interpolation. This is used to update the phase state and grain growth state based on the temperature history and to calculate the microscopic solute concentration distribution. The solute redistribution amount at the solid-liquid interface is determined using the microscopic solute concentration distribution, which is then fed back to the macroscopic FE model to correct the latent heat release and optimize the solution of the billet temperature field distribution. The macroscopic FE model and the microscopic CA model use the same time step to ensure the spatiotemporal synchronization of the macroscopic physical field and the microscopic organizational evolution.

Images