Method and system for optimizing particle size and solidification structure of nickel-based superalloy powder

Abstract

The invention discloses a nickel-based high-temperature alloy powder particle size and solidification structure optimization method, which comprises the following steps of: obtaining a corresponding relation between different particle sizes and cooling rates in a continuous atomization process window through a simulation step, carrying out microscopic structure characterization on powder with different particle size groups through an experiment, and obtaining experimental data of a solidification structure; and coupling results of the simulation step and the experiment step, establishing a quantitative mapping relation model among the particle size, the cooling rate and the solidification assembly, based on the quantitative mapping relation model, taking a target solidification structure as input, carrying out backstepping to obtain target particle size distribution and corresponding atomization process parameters, and adjusting the operation state of equipment in real time through a closed-loop control system. According to the method, through multi-physics field coupling simulation and real-time experimental verification, the quantitative mapping relation among the powder particle size, the cooling rate and the solidification structure is established, and accurate regulation and control of molten drop breaking, cooling and solidification paths in the gas atomization process are achieved.

Description

A method and system for optimizing the particle size and solidification structure of nickel-based superalloy powder Technical Field

[0001] This invention belongs to the field of metal powder preparation technology, specifically relating to a method and system for optimizing the particle size and solidification structure of gas-atomized nickel-based high-temperature alloy powder based on the synergy of computational simulation and experimental characterization. Background Technology

[0002] Nickel-based superalloys are widely used in aerospace, nuclear energy, and gas turbine fields due to their excellent high-temperature strength, oxidation resistance, and corrosion resistance. Gas atomization technology, as a core method for preparing high-quality nickel-based alloy powders, can achieve powders with high sphericity, low oxygen content, and good flowability.

[0003] However, existing gas atomization processes have significant limitations in preparation: the cooling rate of the molten droplets is affected by multiple factors such as particle size, atomizing gas flow field, and droplet trajectory, leading to significant differences in the solidification structure of powders with different particle sizes, resulting in problems such as microstructural heterogeneity, elemental segregation, and inhomogeneous dendrite size. These microstructural inhomogeneities directly affect the densification behavior and mechanical properties of subsequent additive manufacturing or powder metallurgy parts.

[0004] Currently, powder particle size and microstructure control mainly rely on empirical parameter adjustments, such as nozzle pressure, superheat, and gas-liquid ratio, lacking a quantitative guidance mechanism involving multi-physics coupling. Furthermore, traditional experimental methods cannot analyze the complex thermal flux changes during droplet cooling in real time, making it difficult to establish a quantitative mapping relationship between particle size, cooling rate, and solidification microstructure. While existing studies have attempted to introduce CFD or thermal history simulations, these are mostly limited to predicting macroscopic atomization behavior and have not yet achieved predictable control over the solidification microstructure of individual droplets.

[0005] Therefore, there is an urgent need for a systematic approach that combines computational simulation and experimental verification to achieve synergistic control of powder particle size distribution, thermal history, and solidification structure, so as to realize controllable structure and performance feedforward design. Summary of the Invention

[0006] This invention aims to solve the problem that particle size-dependent solidification structure differences cannot be predicted and controlled in the traditional gas atomization powder production process. It proposes a method and system for optimizing the particle size and solidification structure of gas atomized nickel-based superalloy powder based on the synergy of computational thermohydrodynamics (CTFD) simulation and experimental characterization.

[0007] Therefore, the present invention provides a method for optimizing the particle size and solidification structure of nickel-based superalloy powder, comprising the following steps: Step 1, Simulation Step: A transient thermo-fluid-solidification multiphase flow model of atomized droplets is constructed using computational thermo-hydrodynamics to simulate the breakup, cooling, and solidification processes of the droplets under different atomization process parameters. The temperature gradient G and solidification rate R are calculated based on the transient temperature field of the droplets, and the cooling rate V is obtained. By simulating multiple atomization process parameter points, the correspondence between different particle sizes d and cooling rate V within the continuous atomization process window is obtained; Step 2, Experimental Characterization Step: An industrial-grade vacuum induction gas atomization device is used. Prepare nickel-based superalloy powder, keeping the process parameters consistent with the simulation conditions, obtain different particle size groups through graded sieving, and characterize the microstructure of powders in different particle size groups to obtain experimental data on solidification structure; Step 3, mapping relationship model construction step, couples the results of the simulation step and the experimental step to establish a quantitative mapping relationship model between particle size d, cooling rate V, and solidification components; Step 4, feedback control step, based on the quantitative mapping relationship model, takes the target solidification structure as input, back-calculates to obtain the target particle size distribution and corresponding atomization process parameters, and adjusts the equipment operating status in real time through a closed-loop control system.

[0008] Specifically, the quantitative mapping relationship model includes: the relationship between cooling rate and particle size: V = f(d); the relationship between cooling rate and secondary dendrite arm spacing: SDAS = g(V); and the relationship between cooling rate and Laves phase volume fraction: Laves_vol = h(V).

[0009] Specifically, the multiphase flow model uses the VOF method to track the gas-liquid interface and the Realizable k-ε model to simulate turbulence effects. At the same time, the Lagrangian Particle Tracking algorithm is introduced to track the trajectory and thermal history of droplets.

[0010] Specifically, the process parameters set in the simulation step include: atomizing gas pressure of 3.5-6.0 MPa and melt superheat of 100-300 ℃.

[0011] Specifically, in the feedback control step, a machine learning algorithm is used to train and optimize the quantitative mapping relationship model. The machine learning algorithm is support vector regression, and the pressure of the atomized gas or the superheat of the melt is dynamically adjusted by the PLC controller.

[0012] Specifically, the target solidification structure is: suppressing the precipitation of the Laves phase, making the volume fraction of the Laves phase less than 2%; and / or controlling the secondary dendrite arm spacing SDAS to be less than 1.0 μm.

[0013] A solidification microstructure optimization system includes: a gas atomization device for preparing nickel-based superalloy powder under controllable process parameters; a numerical simulation module for performing the computational thermodynamics simulation and outputting droplet size, cooling rate, temperature gradient, and solidification velocity data; a microstructure characterization module for quantitative analysis of the morphology, dendritic structure, and elemental distribution of the prepared powder; and a data optimization and feedback control module, communicatively connected to the numerical simulation module and the microstructure characterization module, configured to: construct and update the quantitative mapping relationship model, optimize process parameters through model inversion based on target microstructure requirements, and issue control commands to the gas atomization device.

[0014] Specifically, the data optimization and feedback control module is integrated with commercial CFD software and characterization equipment software through an API interface, and the gas atomization device is a vacuum induction melting gas atomization device, equipped with a programmable logic controller for precise execution of process parameter adjustments.

[0015] Compared with the prior art, the present invention has the following beneficial effects: For the first time, multiphysics simulation and experimental characterization are applied in combination to gas atomization powder production, establishing a quantitative correspondence between particle size, cooling rate and solidification structure, realizing predictable control of powder structure, effectively controlling dendrite segregation and Laves phase formation among powders of different particle sizes, and significantly improving the uniformity of powder structure and the mechanical consistency of additively manufactured parts. Detailed Implementation

[0016] A method and system for optimizing the particle size and solidification structure of gas-atomized nickel-based superalloy powder based on computational thermohydrodynamics (CTFD) simulation and experimental characterization includes the following steps: Step 1: Simulation Step. A transient heat-fluid-solidification multiphase flow model of gas-atomized droplets is constructed using computational thermohydrodynamics to simulate the breakup, cooling, and solidification processes of molten droplets under different atomization process parameters. Among them, a two-dimensional axisymmetric gas-liquid interface evolution model is constructed, with an initial droplet diameter of 5–60 μm; the atomizing gas pressure is set to 3.5–6.0 MPa; and the superheat is set to 100–300 ℃. The setting methods for atomization parameters and computational domain are as follows: Step A: Determine the reference phase transformation temperature of the material. The liquidus and solidus temperatures of the alloy are determined by referring to literature or experiments.

[0017] For example, the liquidus temperature of the nickel-based superalloy IN718, which is 1600 K, can be used as the initial temperature of the droplets.

[0018] Select the superheat ΔT (depending on the equipment and process): It is recommended to use 100–300 ℃ (≈100–300 K).

[0019] Therefore, the initial droplet temperature in the computational domain is T liq = 1600 + (100~300) K.

[0020] Step B: Calculate the liquid phase (melt) ejection velocity Vliq. Based on the liquid mass flow rate m and the nozzle diameter D, and the law of conservation of mass, determine the gas flow field velocity, which is usually set to 10-50 m / s. This value can be used to calculate and as the boundary inlet velocity (gas flow field velocity) in the simulation.

[0021] The interface was tracked using the VOF method, and the cooling rate distribution was solved by combining the latent heat of phase change with the energy equation. Turbulence was handled using a Realizable k-ε model, and radiative heat transfer was introduced through the DO model. The time step was set to 5 × 10⁻⁶. -7 The minimum grid size was 0.125 μm; the spatial trajectory and residence time of droplets of different sizes were tracked using the Lagrangian Particle Tracking method.

[0022] The temperature gradient G and solidification rate R are calculated based on the transient temperature field of the droplets, and the cooling rate V = f(G, R) is obtained. By simulating multiple atomization process parameter points (pressure, superheat, nozzle diameter, etc.), the correspondence between different particle sizes d and cooling rate V, temperature gradient G, and solidification rate R within a continuous process window is obtained.

[0023] Step 2: Experimental Characterization Steps. Nickel-based superalloy powder was prepared using an industrial-grade vacuum induction gas atomization (VIGA) device, maintaining process parameters consistent with simulation conditions. Different particle size groups were obtained through graded sieving, and the microstructure of the powders in different particle size groups was characterized to obtain experimental data on the solidification structure. Specifically: Different particle size groups (5, 15, 30, 60 μm) were obtained through graded sieving, and the following experimental characterizations were performed: Microstructure analysis, using SEM and EBSD to measure the secondary dendrite arm spacing (SDAS) and orientation characteristics; Elemental segregation characterization, using EDS to quantitatively determine the segregation coefficients of Nb, Ti, Mo, etc.; Phase composition and inclusion analysis, using XRD and TGA to determine the phase composition and oxygen content; Experimental data on the relationship between particle size d and SDAS, degree of segregation, and volume fraction of Laves phase were obtained.

[0024] Step 3: Constructing the mapping relationship model. Couple the results of the simulation and experimental steps to establish a quantitative mapping relationship model between particle size d, cooling rate V, temperature gradient G, and solidification rate R. The model includes the following relationship between particle size and cooling rate: V = f(d), such as: V = −2.1×10⁻⁶. 6 ·d -1.15The relationship between cooling rate and secondary dendrite arm spacing is: SDAS = g(V) or SDAS = 0.197·d 0.518 The functional relationship between cooling rate and segregation degree and Laves phase volume fraction is: Laves_vol = h(V).

[0025] The model is trained using multiple regression or support vector regression (SVR) algorithms and can be used to predict solidification microstructure under arbitrary particle size and process parameters.

[0026] Step 4: Feedback Control Step. Based on the quantitative mapping relationship model, the target solidification structure is used as input to back-calculate the target particle size distribution and corresponding atomization process parameters. The cooling rate V is calculated from the target SDAS, and the particle size distribution d is calculated from V. Based on d, atomization parameters such as atomization pressure and superheat are calculated. The atomization pressure, gas-liquid ratio, melt temperature and flow rate are dynamically adjusted by the PLC control system. The control system records the operating data such as temperature and pressure in real time and automatically fine-tunes the process parameters so that the actual generated powder structure gradually approaches the target structure, realizing closed-loop control.

[0027] This invention is the first to synergistically apply multiphysics simulation and experimental characterization to gas atomization powder production, establishing a quantitative correlation between particle size, cooling rate, and solidification structure, enabling predictable control of powder microstructure. This method effectively suppresses dendritic segregation and Laves phase formation among powders of different particle sizes, significantly improving powder microstructure uniformity and the mechanical consistency of additively manufactured parts. The system possesses good versatility and scalability, and can be used for microstructure optimization of other nickel-based or cobalt-based alloy powders, demonstrating high engineering application value.

[0028] This invention also provides a solidification structure optimization system, comprising: a gas atomization device for preparing nickel-based superalloy powder under controllable process parameters; a numerical simulation module for performing the computational thermodynamics simulation and outputting droplet size, cooling rate, temperature gradient, and solidification velocity data; a structure characterization module for quantitatively analyzing the morphology, dendritic structure, and elemental distribution of the prepared powder; and a data optimization and feedback control module, communicatively connected to the numerical simulation module and the structure characterization module, configured to: construct and update the quantitative mapping relationship model, optimize process parameters through model inversion based on target structure requirements, and issue control commands to the gas atomization device.

[0029] Specifically, the data optimization and feedback control module is integrated with commercial CFD software and characterization equipment software through an API interface, and the gas atomization device is a vacuum induction melting gas atomization device, equipped with a programmable logic controller for precise execution of process parameter adjustments.

[0030] Taking Inconel 718 nickel-based superalloy as a specific example, this invention implements its method and system, detailing the entire process from preparation to optimization. Inconel 718's main composition is Ni-19Cr-18Fe-5Nb-3Mo-1Ti-0.5Al (wt%). Its γ'' and γ'-strengthening properties are crucial for high-temperature performance, but the Laves phase easily forms during gas atomization, leading to embrittlement. This implementation, through a combination of simulation and experimentation, controls particle size distribution to optimize the solidification structure, achieving a powder yield >50%, suitable for laser powder bed fusion (L-PBF) additive manufacturing.

[0031] Step 1: CTFD Numerical Simulation Model Establishment and Droplet Thermal History Calculation. In the modeling and simulation phase, the CTFD method is used to construct a transient thermal-fluid-solidification multiphase flow model of gas-atomized droplets. In ANSYS Fluent 2023 software, a two-dimensional axisymmetric geometric model is constructed with a computational domain size of 240 μm (radial) × 1000 μm (axial). The droplet is placed at the center to avoid boundary effects. Considering an initial droplet diameter range of 5-60 μm, the primary and secondary fragmentation processes of the molten droplet are simulated by a high-pressure argon jet (pressure 3.5-6.0 MPa). The VOF method is used to accurately capture the gas-liquid interface evolution. Combining the energy conservation equation and the latent heat of phase change, the spatiotemporal distribution of the droplet cooling rate is calculated. Through this simulation, the temperature gradient G (K / m) and solidification velocity R (m / s) of droplets with different sizes can be quantified, thereby predicting the solidification morphology, such as the cellular structure formed by a high G / R ratio under small droplet size. Furthermore, a Lagrangian Particle Tracking (LPT) coupled algorithm is introduced to track droplet trajectories, avoiding excessive computational overhead of the VOF method at high Reynolds numbers. Secondly, in parameter setting and boundary condition optimization, for typical nickel-based superalloys such as Inconel 718, a melt superheat of 300 °C and argon gas purity >99.99% are set to minimize oxide inclusions. The nozzle diameter is set to 0.5-1.0 mm, and the cavity vacuum degree is <10. -3 Pa. In the simulation, the mesh was adaptively refined to a minimum element size of 0.125 μm to ensure the interface resolution met the requirements of the Kelvin-Helmholtz instability analysis for droplet breakup. These settings are based on existing gas atomization research, and increasing the atomization pressure can reduce the average particle size D. 50Approximately 20%, but a balance must be struck between energy consumption and powder yield. Boundary conditions also incorporate wall heat flux and convection coefficient to simulate the heat loss of an actual vacuum induction melting-gas atomization (VIGA) device. Through parameter sensitivity analysis, the gas-liquid ratio (10:1-20:1) is optimized to achieve a narrower target particle size distribution, such as a standard deviation σ < 10 μm. Simulation results show that the cooling time for 5 μm droplets is < 1 ms at a rate of 1.8 × 10^6 K / s; the cooling time for 60 μm droplets is > 10 ms at a rate of 5.1 × 10^6 K / s. 5 K / s. The temperature field curve exhibits an exponential decay, with surface cooling faster than the interior, confirming that increased particle size leads to increased heat transfer resistance. Furthermore, G = |∇T| and R = V were calculated. n (Interface normal velocity), G / R distribution reveals the threshold for solidification mode transition.

[0032] Step 2: Preparation and particle size classification of gas-atomized powder. A vacuum induction melting furnace (capacity 50 kg) was selected, and the process parameters were kept consistent with the simulation conditions. Raw materials with a purity >99.9% were batched according to the stoichiometric ratio, and the furnace was evacuated to 10°C. -3 After Pa, argon protection was applied, and the temperature was raised to 1600 °C for 2 hours to ensure uniform composition. The superheat was controlled at 300 °C. The liquid flowed through a bottom pull rod (0.8 mm diameter) into the atomization chamber, and high-purity argon gas (pressure 4.5 MPa, flow rate 5 kg / min) was injected from a coaxial nozzle to break up the droplets. The atomization distance was 1.2 m. After collecting the powder, it was classified into four groups using a vibrating sieve: 5 μm (<10 μm), 15 μm (10-20 μm), 30 μm (20-40 μm), and 60 μm (40-80 μm). The particle size was measured using a particle size analyzer. 50 The particle sizes were 8.9, 18.7, 33.5, and 61.2 μm, respectively, and the distribution was log-normal. This step ensured sample representativeness and avoided particle size overlap affecting subsequent analysis. The powder oxygen content was <150 ppm as determined by TGA, and the sphericity was >96% as calculated by image processing software. No obvious oxide scale or inclusions were found.

[0033] Step 3: Characterization of solidification structure and dendrite analysis to obtain experimental data on the relationship between particle size d and SDAS, degree of segregation, and volume fraction of Laves phase. After embedding the graded powder into resin and polishing, the surface morphology was observed by SEM (5000-20000× magnification): small particles were smooth without satellites, while large particles occasionally showed fine adhesions. The maximum EBSD scanning area was 50×50 μm with a resolution of 0.1 μm, showing crystal orientation and low-angle grain boundaries (>2°). 5 μm powder exhibited equiaxed cellular crystals, SDAS=0.5 μm, and texture intensity <2 times randomness; the primary dendrite arm length of 30 μm powder was 1.1 μm. <111> Orientation-dominant; 60 μm powder SDAS = 1.64 μm, well-developed secondary arms, Laves phase volume fraction 3.2%. EDS line scanning (0.5 μm step size) showed Nb / Ti enriched 1.4 times at grain boundaries, Mo depleted 0.85 times, and segregation intensified with increasing SDAS. An empirical formula SDAS = 0.197 d was established based on 1000 SDAS values. 0.518 This analysis verified the simulation accuracy, with a deviation of <10%, revealing enhanced interface stability and suppression of dendrite coarsening under high cooling rates.

[0034] Step 4: Couple the results of the simulation and experimental steps to establish a quantitative mapping model between particle size d, cooling rate V, and solidification components, based on the simulation result V = −2.1×10⁶·d. -1.15 Based on the experimental SDAS data, SDAS = 0.197·d 0.518 A regression model was constructed to predict the solidification path; combined with simulated temperature gradients G (10⁵–10⁷ K / m) and R (0.01–1 m / s), a solidification map was plotted: in the high G / R region (>10⁶), planar / cellular crystals form, while in the low G / R region (<10⁴), dendrites dominate. For Inconel 718, the Scheil-Gulliver model predicted the Nb segregation coefficient k. Nb =0.85, at low cooling rates, Laves (Ni,Nb,Fe,Cr)₂(Laves) phase precipitates, reducing ductility by 20%. In the experiment, 5 μm powder had no Laves phase and a hardness HV=350; 60 μm powder showed enhanced Laves phase, with a hardness HV=420, but increased brittleness. Mechanistically, small particle size and high cooling rates shorten the solid-liquid coexistence time (<10). -4} s), reducing solute diffusion, segregation coefficient <1.1; large particle size and low cooling rate (10 5 (K / s) Extending the diffusion path leads to Laves formation. Model validation using error sum-of-squares minimization confirms that particle size-controlled cooling rate is key to suppressing heterogeneity. Furthermore, phase-field simulation is introduced to predict the evolution of the microscopic solute field, supporting macro-microscale bridging.

[0035] Step 5: System Integration and Process Optimization Feedback. The mapping model is embedded into the optimization system, and simulation / experimental data is processed using the MATLAB interface to construct an inversion algorithm: when the target SDAS < 1.0 μm, feedback is provided to reduce the particle size (increase gas pressure to 4.5 MPa) or increase the superheat to 150 °C. In practical applications, the initial process D... 50 =45 μm, SDAS=1.5 μm; Optimized D 50 =25 μm, SDAS=0.8 μm, Laves phase <1%. The system monitors infrared thermometer data in real time and compares it with simulation. The PID controller adjusts the air pressure to adjust the airflow to 80-100 m / s, improving the yield by 15%. For additive manufacturing, the optimized powder L-PBF formed parts have a density >99.5%, tensile strength >1200 MPa, and uniformity 10% better than commercial powders. Extended verification shows that similar patterns hold true when applied to René 104 alloy, confirming its versatility. Through the above implementation, the method of this invention effectively controls the microstructure of Inconel 718 powder, and the quantitative relationship between cooling rate and SDAS lays the foundation for alloy design, possessing significant engineering value.

Claims

1. A method for optimizing the particle size and solidification structure of nickel-based superalloy powder, characterized in that, Includes the following steps: Step 1: Simulation Step. A transient thermo-fluid-solidification multiphase flow model of atomized droplets is constructed using computational thermo-hydrodynamics. This model simulates the breakup, cooling, and solidification processes of the droplets under different atomization process parameters. The temperature gradient G and solidification rate R are calculated based on the transient temperature field of the droplets, and the cooling rate V is obtained. By simulating multiple atomization process parameter points, the correspondence between different particle sizes d and the cooling rate V within the continuous atomization process window is obtained. Step 2: Experimental Characterization Step. Nickel-based superalloy powder is prepared using an industrial-grade vacuum induction atomization device. The process parameters are kept consistent with the simulation conditions. Different particle size groups are obtained through graded sieving, and the microstructure of the powders in different particle size groups is characterized to obtain experimental data on the solidification structure. Step 3: Mapping Relationship Model Construction Step. The results of the simulation and experimental steps are coupled to establish a quantitative mapping relationship model between particle size d, cooling rate V, and solidification components. Step 4: Feedback Control Step. Based on the quantitative mapping relationship model, the target particle size distribution and corresponding atomization process parameters are obtained by back-calculation using the target solidification structure as input. The equipment operating status is adjusted in real time through a closed-loop control system.

2. The method for optimizing the particle size and solidification structure of nickel-based superalloy powder according to claim 1, characterized in that, The quantitative mapping relationship model includes: the relationship between cooling rate and particle size: V = f(d); the relationship between cooling rate and secondary dendrite arm spacing: SDAS = g(V); and the relationship between cooling rate and Laves phase volume fraction: Laves_vol = h(V).

3. The method for optimizing the particle size and solidification structure of nickel-based superalloy powder according to claim 1, characterized in that, The multiphase flow model uses the VOF method to track the gas-liquid interface and the Realizable k-ε model to simulate turbulence effects. At the same time, the Lagrangian Particle Tracking algorithm is introduced to track the trajectory and thermal history of droplets.

4. The method for optimizing the particle size and solidification structure of nickel-based superalloy powder according to claim 1, characterized in that, In the simulation step, the set process parameters include: atomizing gas pressure of 3.5-6.0 MPa and melt superheat of 100-300 ℃.

5. The method for optimizing the particle size and solidification structure of nickel-based superalloy powder according to claim 4, characterized in that, In the feedback control step, a machine learning algorithm is used to train and optimize the quantitative mapping relationship model. The machine learning algorithm is support vector regression, and the pressure of the atomized gas or the superheat of the melt is dynamically adjusted by the PLC controller.

6. The method for optimizing the particle size and solidification structure of nickel-based superalloy powder according to claim 1, characterized in that, The target solidification structure is: suppressing the precipitation of the Laves phase, making the volume fraction of the Laves phase less than 2%; and / or controlling the secondary dendrite arm spacing SDAS to be less than 1.0 μm.

7. A coagulation structure optimization system for implementing the method of any one of claims 1-6, characterized in that, include: A gas atomization device is used to prepare nickel-based superalloy powders under controllable process parameters; The numerical simulation module is used to perform the computational thermodynamics simulation and output data on droplet size, cooling rate, temperature gradient and solidification rate. The tissue characterization module is used for quantitative analysis of the morphology, dendritic structure and elemental distribution of the prepared powder; the data optimization and feedback control module is communicatively connected to the numerical simulation module and the tissue characterization module, and is configured to: construct and update the quantitative mapping relationship model, optimize process parameters by inverting the model based on the target tissue requirements, and send control commands to the gas atomization device.

8. The system according to claim 7, characterized in that, The data optimization and feedback control module is integrated with commercial CFD software and characterization equipment software through an API interface. The gas atomization device is a vacuum induction melting gas atomization device and is equipped with a programmable logic controller for precise execution of process parameter adjustments.