A simulation method for high-cycle fatigue of turbine rotors based on fluid-structure interaction

Abstract

This invention belongs to the field of turbine machinery fatigue simulation and reliability assessment technology, and relates to a high-cycle fatigue simulation method for turbine rotors based on fluid-structure interaction. First, a three-dimensional model of the fluid and solid domains is constructed based on the actual turbine model. Second, transient flow field simulation is performed to obtain turbine aerodynamic loads, and the aerodynamic loads are verified by analyzing their frequency domain characteristics. Third, transient structural simulation analysis is conducted, mapping aerodynamic loads at different times onto the turbine solid structure mesh nodes, and iterating the damping by comparing with experimental data to output the transient stress field of the turbine surface nodes. Finally, transient fatigue life simulation is performed, outputting the maximum principal stress at each time point of the turbine's key nodes, and the turbine rotor life is estimated based on the stress-life method.

Description

Technical Field

[0001] This invention belongs to the field of turbine machinery fatigue simulation and reliability assessment technology, specifically involving a simulation method for accurately predicting high cycle fatigue (HCF) of turbine rotors based on fluid-structure interaction under extreme operating conditions such as aero-engines and gas turbines. Background Technology

[0002] Aero engines and gas turbines operate under extreme combined load environments of "high temperature, high stress, and high frequency" for extended periods. On one hand, high-speed rotation subjects the blades to a state of continuous high stress; on the other hand, the dynamic-static interference effect between the stator and rotor generates high-frequency aerodynamic excitation, inducing dynamic aerodynamic stress. When these two stresses are superimposed, even if the stress level is below the material's yield strength, hundreds of millions of cycles can still lead to high-cycle fatigue failure of the blades. The resulting fragments, carried by the high-speed airflow, can impact subsequent components, causing catastrophic safety accidents.

[0003] Current turbine high-cycle fatigue simulation technology faces two major bottlenecks: 1. Low accuracy of stress superposition: Centrifugal stress is usually calculated in structural simulation software (such as Mechanical), while aerodynamic stress is obtained in fluid simulation software (such as Fluent). The two are based on different grid systems and the grids are in constant motion. Manual interpolation conversion is required when superimposing the stress, resulting in a superposition error of more than 15%, which cannot accurately reflect the actual stress state of the blade. 2. Poor flexibility in fatigue calculation: Existing fatigue life calculations mostly rely on integrated modules such as ANSYS-Ncode, which limits the adjustment of frequency domain analysis parameters and fatigue correction models (such as Goodman correction), making it difficult to adapt to different turbine models and material properties, and the life prediction deviation can reach more than 30%.

[0004] Furthermore, traditional design methods rely excessively on empirical formulas and bench tests, leading not only to low material utilization (with safety factors typically set at 1.5-2.0), which restricts the improvement of engine thrust-to-weight ratio, but also to high testing costs (millions of yuan per test) and long cycles (3-6 months), making it difficult to meet the needs of rapid iterative turbine design. Therefore, there is an urgent need for a simulation method that balances the accuracy of stress superposition with the flexibility of fatigue calculation to solve the problems of inaccuracy and inefficiency in high-cycle fatigue prediction of turbine rotors. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies. This invention provides a high-cycle fatigue simulation method for turbine rotors based on fluid-structure interaction. By virtually rotating and matching dynamic fluid meshes with static solid meshes, aerodynamic loads are accurately interpolated onto the turbine structure mesh using the basis function method and the transfer matrix method. Furthermore, the high-cycle fatigue SN data is corrected based on Matlab, taking into account various processing factors and the influence of SN extrapolation to the ultra-high cycle region, thereby improving the accuracy of fatigue calculation.

[0006] The technical solution of this invention: A high-cycle fatigue simulation method for turbine rotors based on fluid-structure interaction is proposed. First, a three-dimensional model of the fluid and solid domains is constructed based on an actual turbine model. Second, transient flow field simulation is performed to obtain turbine aerodynamic loads, and the frequency domain characteristics of the aerodynamic loads are analyzed to verify the loads. Third, transient structural simulation analysis is conducted, mapping aerodynamic loads at different times onto the turbine solid structure mesh nodes, and iterating the damping by comparing with experimental data to output the transient stress field of the turbine surface nodes. Finally, transient fatigue life simulation is performed, outputting the maximum principal stress at each time point of the turbine's key nodes, and the turbine rotor life is estimated based on the stress-life method.

[0007] Furthermore, the actual turbine model is a simplified three-dimensional simulation model that includes a turbine guide vane and a turbine rotor. Redundant components that have no effect on the flow field inside the turbine are removed, and the simplified three-dimensional simulation model is divided into a fluid domain and a solid domain. The fluid domain includes a guide vane flow domain and a rotor flow domain. The guide vane flow domain is set as a stationary domain, and the rotor flow domain is set as a rotating domain. The solid domain includes the turbine rotor structure. An interface is set between the fluid domain and the solid domain.

[0008] Furthermore, transient flow field simulations were conducted to obtain turbine aerodynamic loads. The aerodynamic loads were verified by analyzing their frequency domain characteristics, including: Mesh the fluid domain; The generated mesh was imported into the pressure-based transient solver of the transient flow field simulation software Fluent. The medium, turbulence model, total pressure at the inlet boundary, total temperature at the inlet boundary, static pressure at the outlet boundary, interface between the guide flow domain and the rotor flow domain, MRF rotating coordinate system and rotational speed, and calculation step size were set in the pressure-based transient solver. Extract aerodynamic loads. When there is a clear periodicity in the pressure on the blade surface, extract the aerodynamic loads on the turbine surface. According to a certain calculation step size, output the coordinates (X,Y,Z) of the center of the turbine surface grid and the static pressure P respectively to obtain sample data at multiple different times. Generate a "number-fluid grid point coordinates-aerodynamic load" file containing a time series within a period from the sample data. Frequency domain analysis was performed on the pressure-time signals from sample data at different times to analyze whether the peak frequency of energy concentration corresponds to the nozzle passage frequency. and harmonics Blade passing frequency and its harmonics Similarly, we will verify whether the aerodynamic load on the turbine surface matches the theoretical formula and simulation results.

[0009] Furthermore, during the mesh generation of the fluid domain, the mesh on the turbine blade surface is locally refined to ensure that the number of mesh layers at the 0.096mm gap between the blade and the wall is ≥3, the minimum orthogonal mass of the mesh is ≥0.1, and the first boundary layer Y+≈1 on the turbine blade surface.

[0010] Furthermore, the medium in the pressure-based transient solver was set as an ideal gas, and the SST k-omega turbulence model was selected. The inlet boundary was set as a pressure inlet with a total pressure of 950 kPa and a total temperature of 62 °C, and the outlet boundary was set as a pressure outlet with a static pressure of 127 kPa. The interface between the guide flow domain and the rotor flow domain was set as a sliding mesh. The MRF rotating coordinate system method was used to apply a rotational speed of 90,000 rpm around the +Z axis, and the calculation time step was set to 5*10. -7 s; The isobaric specific heat capacity of the ideal gas is Cp = 1006.43 J / (kg·K), thermal conductivity is 0.0242 W / (m·K), and dynamic viscosity is 1.7894 × 10⁻⁶. -5 kg / (m·s), molecular weight 28.966 kg / kmol.

[0011] Furthermore, transient structural simulation analysis includes: Draw the turbine solid mesh and export the turbine solid surface mesh nodes: Draw the turbine solid mesh through curvature control. The tetrahedral mesh of the aerodynamic load solid domain is refined at the blade tip root. The solid mesh is similar in size to the turbine surface mesh of the fluid domain. Export the coordinates of the solid mesh points. The fluid mesh and solid mesh are mapped one-to-one using the virtual rotation method. That is, the rotating fluid mesh points are mapped one-to-one with the non-rotating solid mesh points by mesh number, and a "number-solid mesh node coordinate" file is generated. A time-series aerodynamic excitation load is generated using bilinear interpolation and transfer matrix methods: The initial aerodynamic load text files "number-fluid grid point coordinates-aerodynamic load" and "number-solid grid node coordinates" are read. Interpolation is performed based on bilinear interpolation to generate an aerodynamic excitation load text file "solid grid point coordinates-aerodynamic excitation". The transfer matrix between the aerodynamic load field and the interpolated solid aerodynamic excitation is calculated. The aerodynamic excitation at all subsequent times is calculated using this transfer matrix to generate a time-series aerodynamic excitation load text file "time-solid grid point coordinates-aerodynamic excitation". Set up the ANSYS Transient Structure calculation module. In this calculation module, set the turbine material, constraints and loads, import the "Time-Solid Mesh Point Coordinates-Aerodynamic Excitation" file, and input the aerodynamic excitation load into the calculation module to calculate the transient stress peak. Compare the calculated transient stress peak with the experimental data, and adjust the damping to ensure that the damping falls within a reasonable range; Output turbine stress field history data, i.e., output a file of "turbine surface grid point coordinates-stress-time".

[0012] Furthermore, in the ANSYS Transient Structure calculation module, the turbine material is set to 2A12 aluminum alloy; Constraints and load application: The inner cylindrical surface of the turbine is set as a Cylindrical Support to restrict radial and axial displacement, while allowing tangential motion; a rotational speed of 90,000 rpm is applied.

[0013] Furthermore, transient fatigue life simulation is performed, including: Read the turbine mesh node coordinates and stress history data from the "Turbine Surface Mesh Point Coordinates-Stress-Time" file; Rainflow calculations were performed on the stress history of each grid point on the turbine surface to obtain the stress cycle amplitude and average stress corresponding to each grid point on the turbine surface. The stress amplitude and average stress of an asymmetric cycle are equivalent to the stress amplitude of a symmetric cycle. By fitting the SN data in the high-cycle region of the turbine, SN curves for different turbine survival rates are constructed; Estimate the lifespan of each grid point on the turbine.

[0014] Furthermore, by fitting the SN data in the high-cycle region of the turbine, SN curves for different turbine survival rates were constructed, including: The SN data of the high-frequency region of 2A12 aluminum alloy were fitted using the Basquin formula; Introducing the effect of surface roughness Surface treatment and correction Residual stress correction and size effect factor Correct the SN data; The fatigue life of 2A12 aluminum alloy is assumed to follow a log-normal distribution, and the logarithmic standard deviation of fatigue life is 0.3 under the same stress level. SN curves corresponding to survival rates of 90%, 99%, and 99.9% are constructed.

[0015] Furthermore, the lifespan of each grid point of the turbine is estimated using Miner's linear damage accumulation rule: Let the theoretical fatigue life of the i-th stress cycle at the corresponding stress level be... The damage caused by this cycle is The total damage D is obtained by accumulating the damage from each cycle. Turbine grid point lifespan .

[0016] The beneficial effects of this invention are as follows: This invention provides a high-cycle fatigue simulation method for turbine rotors based on fluid-structure interaction. By virtually rotating and matching the dynamic fluid mesh and the static solid mesh, the aerodynamic load is accurately interpolated onto the turbine structure mesh using the basis function method and the transfer matrix method. Furthermore, the high-cycle fatigue SN data is corrected based on Matlab, taking into account various processing factors and the influence of SN extrapolation to the ultra-high cycle region, thereby improving the accuracy of fatigue calculation. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a structural diagram of the original model of the turbine assembly; Figure 3 This is a schematic diagram of the turbine fluid domain; Figure 4 This is a schematic diagram of the turbine fluid domain mesh. Figure 5 The image is a txt file showing the aerodynamic loads on the turbine surface. Figure 6 The results of the frequency domain analysis of aerodynamic loads are shown in the figure. Figure 7 Mesh diagram of the turbine solid domain; Figure 8 This is an aerodynamic excitation diagram generated based on solid mesh interpolation; Figure 9 Stress history diagram for transient structural analysis of aerodynamic excitation; Figure 10 Ultra-high frequency SN curves corrected for turbine materials Figure 1 ; Figure 11 Ultra-high frequency SN curves corrected for turbine materials Figure 2 ; Figure 12 Output the critical points and critical point life diagrams of the turbine blades for Matlab. Detailed Implementation

[0018] The following description of the embodiments further illustrates the specific implementation of the present invention in detail, so as to help those skilled in the art to have a more complete, accurate and in-depth understanding of the concept and technical solution of the present invention: One embodiment of the present invention provides a method for simulating high-cycle fatigue of turbine rotors based on fluid-structure interaction, comprising the following steps: 1) Create turbine fluid and solid domain simulation models: Establish a simplified 3D simulation model including turbine guide vanes and rotors, remove redundant components that have no effect on the internal flow field of the turbine, and divide the simplified 3D simulation model into fluid and solid domains. The fluid domain includes the guide vane flow domain and the rotor flow domain. The guide vane flow domain is set as a stationary domain, and the rotor flow domain is set as a rotating domain. The solid domain includes the turbine rotor structure. An interface is set between the fluid and solid domains, and the simulation models do not need to share topology.

[0019] 2) Conduct transient flow field simulation: ① Fluid Domain Meshing: ANSYS Meshing tool was used to generate a polyhedral mesh with a total of 3.25 million elements. The mesh on the blade surface was locally refined (size 0.1-0.3mm), with a focus on ensuring that the number of mesh layers at the 0.096mm gap between the blade and the wall was ≥3 (to meet the requirements of boundary layer flow calculation). Mesh quality checks showed a minimum orthogonality quality of 0.21, which meets the accuracy requirements for numerical calculations. The thickness of the first boundary layer on the turbine blade surface was 10. -5 mm, a total of 20 layers, wall surface Y+≈1.

[0020] ② Build a transient CFD module in Fluent: Import the generated mesh into Fluent and set the medium to an ideal gas (isobaric specific heat capacity Cp = 1006.43 J / (kg)). K), thermal conductivity 0.0242 W / (m K), dynamic viscosity 1.7894×10 -5 kg / (m (s), molecular weight 28.966 kg / kmol), using the SST k-omega turbulence model, with the inlet boundary set as a pressure inlet of 950 kPa and 62 °C, and the outlet boundary set as a pressure outlet of 127 kPa. The interface between the guide flow domain and the rotor flow domain was set as a sliding mesh. The MRF rotating coordinate system method was used to apply a rotational speed of 90,000 rpm around the +Z axis, and the calculation time step was set to 5*10. -7 s; ③ Extract aerodynamic loads: After the turbine has rotated 10 revolutions (approximately 0.0067s), and a clear periodic pattern is observed in the pressure on the blade surface, the aerodynamic loads on the turbine surface are extracted every 2 time steps (i.e., 10 s). -6s) Output the coordinates (X,Y,Z) of the cellcenter of the turbine surface mesh and the static pressure (P) data, and generate a time series file containing a period of "number-fluid mesh point coordinates-aerodynamic load" (670 files in total, named FLTG-0.006700s.txt, FLTG-0.006701s.txt, FLTG-0.006702s.txt, ..., FLTG-0.007370s.txt).

[0021] ④ Verify aerodynamic load: Perform a fast Fourier transform on the pressure-time signal extracted from the CFD, analyze the peak frequency of energy concentration, and determine whether it corresponds to the nozzle passage frequency. and harmonics Blade passing frequency and its harmonics Similar to the theoretical formula and experimental results, the aerodynamic load on the turbine surface is verified to match the experimental results.

[0022] 3) Conduct transient structural simulation: ① Draw solid mesh and export turbine solid surface mesh nodes: Draw turbine solid mesh through curvature control. The aerodynamic load solid domain tetrahedral mesh is used to refine the blade tip root (stress concentration area). The mesh size must be less than 0.1mm to ensure that the solid mesh is similar in size to the turbine surface mesh in the fluid domain. Export the mesh node coordinates "solid mesh point coordinates". ② The fluid dynamic mesh and the solid mesh are mapped one-to-one by virtual rotation method: In transient flow field simulation, the fluid mesh rotates in reality over time. In transient structure simulation, it is not desirable for the solid mesh to change. By using virtual rotation method, the rotating fluid mesh points and the non-rotating solid mesh points are mapped one-to-one by mesh number, and a file of "number-solid mesh node coordinates" is generated. ③ Generation of time-series aerodynamic excitation loads using bilinear interpolation and transfer matrix methods: To achieve accurate mapping of aerodynamic loads to turbine surface grid nodes, a MATLAB program was written to read the initial aerodynamic load text files "number-fluid grid point coordinates-aerodynamic load" and "number-solid grid node coordinates". Based on the bilinear interpolation method, interpolation was performed to generate the aerodynamic excitation load text file "solid grid point coordinates-aerodynamic excitation". This avoids the defect of "smoothing pressure peaks" in linear interpolation. The transfer matrix between the aerodynamic load field and the interpolated solid aerodynamic excitation was calculated. The aerodynamic excitation at all subsequent times was quickly calculated using this transfer matrix, generating the time-series aerodynamic excitation load text file "time-solid grid point coordinates-aerodynamic excitation". ④ Set up the ANSYS Transient Structure calculation module: Set the turbine material to 2A12 aluminum alloy; Constraints and load application: Set the inner cylindrical surface of the turbine to Cylindrical Support, restricting radial and axial displacement, and allowing tangential motion; Apply a rotational speed of 90,000 rpm; Import "Time-Solid Mesh Point Coordinates-Aerodynamic Excitation" to input the aerodynamic excitation load; ⑤ Iterative Damping: After the calculation is completed, the damping is adjusted by comparing the calculated transient stress peak with the experimental data so that the damping falls within a reasonable range; ⑥ Output turbine stress field history data: Output a file of "turbine surface grid point coordinates-stress-time", which will be used as the load input for subsequent fatigue simulation.

[0023] 4) Conduct transient fatigue life simulation: ① Turbine structure stress data reading and preprocessing: Read the turbine grid node coordinates and stress history data from the "Turbine Surface Grid Point Coordinates-Stress-Time" file; ② Rainflow calculation: Perform rainflow calculation on the stress history of each node to obtain the stress cycle amplitude and average stress corresponding to each grid point on the turbine surface.

[0024] ③ Mean stress correction: Based on Goodman, the stress amplitude and mean stress of asymmetric cycles are equivalent to the stress amplitude of symmetric cycles.

[0025] ④ Fit the SN data of the high-cycle region of the turbine and construct SN curves for different turbine survival rates: The influence of stress-life data extended to the ultra-high cycle region is considered by the probabilistic SN method. Specifically, the SN data of the high-cycle region is fitted by the Basquin formula, taking into account the influence of surface roughness, surface treatment correction, residual stress correction, and size effect factor. Furthermore, considering that the fatigue data follows a log-normal distribution, and considering that the standard deviation of life is 0.3 under the same stress level, SN curves for different survival rates such as 90%, 99%, and 99.9% are constructed. ⑤ Use Miner's linear damage accumulation rule to estimate the life of each grid point of the turbine, that is, calculate the life of each grid point by the damage caused by each cycle and the total damage.

[0026] Specifically, the method for extracting aerodynamic loads in step 2) includes: The method for generating time-series aerodynamic excitation loads using bilinear interpolation and transfer matrix methods in step 3) specifically includes: The initial aerodynamic load text files "Number-Fluid Grid Point Coordinates-Aerodynamic Load" and "Number-Solid Grid Node Coordinates" are read. Interpolation is performed based on the bilinear interpolation method: ① Define the source point (fluid grid face center point) and the target point (structural grid node); ② Lock the grid node number by virtual rotation; ③ Construct the interpolation matrix and add a linear polynomial to ensure the conservation of force and moment; ④ Solve for the interpolation coefficients; ⑤ Interpolate the fluid pressure onto the structural grid based on the interpolation coefficients to generate the aerodynamic excitation load text file "Solid Grid Point Coordinates-Aerodynamic Excitation", avoiding the defect of "smoothing pressure peaks" in linear interpolation.

[0027] The transfer matrix of the above aerodynamic load field and the interpolated solid aerodynamic excitation is calculated. The virtual rotation method is used to correspond the aerodynamic excitation coordinates with the stationary solid grid nodes one by one by the grid node numbering, so as to realize the virtual rotation of the solid grid and its actual stationary state. The transfer matrix is ​​used to quickly calculate the aerodynamic excitation at all subsequent moments.

[0028] Step 4) The Matlab fatigue life calculation includes various corrections, such as surface roughness correction. Based on the actual turbine machining accuracy (Ra 1.6 μm), a reduction factor of 0.85-0.92 is applied to the SN curve to correct the material fatigue performance under actual working conditions. This includes considering the influence of extending the SN curve to the ultra-high cycle fatigue region through a probabilistic method.

[0029] The second embodiment of the present invention refers to Figure 1 The diagram shows a high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction, comprising the following steps: Step 1: Create simulation models of the turbine fluid and solid domains Taking a 75mm diameter turbine as the research object, the core model includes 34 guide vane blades and 6 rotor blades. The model is divided into a fluid domain and a solid domain. In the fluid domain, only the guide vane domain and rotor domain, which have a significant impact on the flow field inside the turbine, are retained. The guide vane domain is a fixed domain, and the rotor domain is a rotating domain. The shell domain, whose influence is negligible, is removed, reducing the number of meshes by 50% and lowering the computational cost. In the solid domain, only the turbine rotor structure is included to ensure that the structural simulation focuses on the core stress-bearing components.

[0030] Step 2: Conduct transient flow field simulation to obtain aerodynamic loads 1) Fluid Domain Meshing: ANSYS Meshing tool was used to generate a polyhedral mesh with a total of 3.25 million elements. The mesh on the blade surface was locally refined (size 0.1-0.3mm), with a focus on ensuring that the number of mesh layers at the 0.096mm gap between the blade and the wall was ≥3 (to meet the requirements of boundary layer flow calculation). Mesh quality checks showed a minimum orthogonality quality of 0.21, which meets the accuracy requirements for numerical calculations. The thickness of the first boundary layer on the turbine blade surface was 10. -5mm, a total of 20 layers, wall surface Y+≈1.

[0031] 2) Set up a transient CFD calculation module in Fluent: Select the pressure-based transient solver in Fluent software; set the medium as an ideal gas (isobaric specific heat capacity Cp = 1006.43 J / (kg)). K), thermal conductivity 0.0242 W / (m K), dynamic viscosity 1.7894e-0.5 kg / (m (s), molecular weight 28.966 kg / kmol); The SST k-omega turbulence model was selected, with the inlet boundary set as a pressure inlet of 950 kPa and 62 °C, and the outlet boundary set as a pressure outlet of 127 kPa. The interface between the guide flow domain and the rotor flow domain was set as a sliding mesh. The MRF rotating coordinate system method was used to apply a rotational speed of 90,000 rpm around the +Z axis, and the calculation time step was set to 5*10. -7 s.

[0032] 3) Extracting aerodynamic loads: After the turbine rotates 10 revolutions (approximately 0.0067s), the aerodynamic loads on the turbine surface are extracted. Every 2 time steps (i.e., 10-6s), the coordinates (X,Y,Z) of the center of the turbine surface grid cell and the static pressure (P) data are output, generating a "number-fluid grid point coordinates-aerodynamic load" file containing a time series within a cycle.

[0033] 4) Verification of aerodynamic loads: Frequency domain analysis was performed on certain characteristic points in 670 sample data at different times. The peaks with the most concentrated energy in the spectrum appeared at 9137Hz, 12182.7Hz, 25888Hz, and 50761Hz, which is consistent with the theoretical formula calculation. =9000Hz, blade-nozzle interference frequency 12750Hz, =25500Hz, =51000Hz, the simulation results are in agreement with the theoretical calculation results.

[0034] Step 3: Conduct transient structural simulation to obtain stress field history 1) Draw solid mesh and export turbine solid surface mesh nodes: Draw turbine solid mesh through curvature control. The aerodynamic load solid domain tetrahedral mesh is used to refine the blade tip root (stress concentration area). Export mesh node coordinates "solid mesh point coordinates".

[0035] 2) Using the virtual rotation method to map the fluid dynamic mesh to the solid mesh: In transient flow field simulation, the fluid mesh rotates in reality over time. In transient structure simulation, it is not desirable for the solid mesh to change. By using the virtual rotation method, the rotating fluid mesh points are mapped to the non-rotating solid mesh points through mesh numbering.

[0036] 3) Generation of time-series aerodynamic excitation loads using bilinear interpolation and transfer matrix methods: To achieve accurate mapping of aerodynamic loads to turbine surface grid nodes, a MATLAB program was written to read the initial aerodynamic load text files "number-fluid grid point coordinates-aerodynamic load" and "number-solid grid node coordinates". Based on the bilinear interpolation method, interpolation was performed to generate the aerodynamic excitation load text file "solid grid point coordinates-aerodynamic excitation", avoiding the defect of "smoothing pressure peaks" in linear interpolation.

[0037] Calculate the transfer matrix of the aerodynamic load field and the interpolated solid aerodynamic excitation. Use this transfer matrix to quickly calculate the aerodynamic excitation at all subsequent times and generate a time series aerodynamic excitation load text file of "time-solid grid point coordinates-aerodynamic excitation".

[0038] 4) Set up the ANSYS Transient Structure calculation module: Set the turbine material to 2A12 aluminum alloy (elastic modulus 71GPa, Poisson's ratio 0.3, density 2.78g / cm³); Constraints and load application: Set the inner cylindrical surface of the turbine to CylindricalSupport, restrict radial and axial displacement, and allow tangential motion; Apply a rotational speed of 90,000 rpm; Import "Solid Mesh Point Coordinates - Aerodynamic Stress - Time" to input the aerodynamic excitation load.

[0039] 5) Iterative damping: After the calculation is completed, the damping is adjusted according to the calculated transient stress peak value and experimental data so that the damping falls within a reasonable range.

[0040] 6) Output turbine stress field history: Output a file of "turbine surface grid point coordinates-stress-time".

[0041] Step 4: Conduct transient lifetime simulation 1) Turbine structure stress data reading and preprocessing, reading turbine mesh node coordinates and stress history data; 2) Rainflow calculation: Perform rainflow calculation on the stress history of each node to obtain the amplitude of the stress cycle corresponding to each grid point. and mean .

[0042] 3) Mean stress correction: Based on Goodman's method, the stress amplitude and mean of asymmetric cycles are equivalent to the stress amplitude of symmetric cycles. .

[0043] 4) The influence of extending the stress-life data of 2A12 aluminum alloy to the ultra-high cycle region is considered using the probabilistic SN method, specifically through the Basquin formula. Fitting SN data in the high-frequency region, considering the influence of surface roughness. Surface treatment and correction Residual stress correction and size effect factor Furthermore, considering that the fatigue data follows a log-normal distribution, and that the standard deviation of the life is 0.3 under the same stress level, SN curves for different survival rates such as 90%, 99%, and 99.9% are constructed. 5) Estimate the life of each grid point of the turbine using Miner's linear damage accumulation rule: Let the theoretical fatigue life of the i-th stress cycle at the corresponding stress level be... The damage caused by this cycle is The total damage D is obtained by accumulating the damage from each cycle. Turbine grid point lifespan .

[0044] Compared with existing technologies, this invention has the following innovative features: 1. Reduced data volume for aerodynamic load storage and improved interpolation accuracy using bilinear interpolation: The fluid domain is rendered using a polyhedral mesh. Compared to the traditional tetrahedral mesh, the polyhedral mesh reduces the number of meshes by 30% and shortens CFD calculation time by more than 40% while maintaining the same flow field calculation accuracy. Furthermore, it extracts load data from the center of the fluid mesh cells using a face-centered rather than node-centered approach, reducing the data volume by more than 50% while ensuring sufficient accuracy. The bilinear interpolation method avoids the "smoothing of pressure peaks" defect of linear interpolation.

[0045] 2. Virtual Rotation Method and Transfer Matrix Method for Generating Time-Series Aerodynamic Excitation Loads: The coordinates of discrete points of the aerodynamic excitation load change over time. To improve the efficiency of transient structure simulation and avoid errors caused by multiple interpolations, a virtual rotation method is used on the solid mesh. The dynamically rotating aerodynamic excitation coordinates are mapped one-to-one with the non-rotating solid mesh points through mesh numbering. Although the solid mesh is stationary during calculation, it achieves virtual rotation. The transfer matrix between the aerodynamic load field and the interpolated solid aerodynamic excitation is calculated. This transfer matrix is ​​then used to quickly calculate the aerodynamic excitation at all subsequent time points, generating a time-series aerodynamic excitation load text file of "time - solid mesh point coordinates - aerodynamic excitation". Calculating the transfer matrix between the above aerodynamic load field and the interpolated solid aerodynamic excitation, and using this transfer matrix to quickly calculate the aerodynamic excitation at all subsequent time points, transforms the aerodynamic load into an aerodynamic excitation, saving 90% of Matlab computation time.

[0046] 3. Quantitative, integrated and automatically executed fatigue life analysis method: Using tools such as Matlab, the execution process of finite element result reading, stress correction, rainflow counting, damage calculation and life output is realized. Through multi-parameter and probabilistic fatigue life calculation, the prediction of turbine ultra-high cycle fatigue life is realized.

[0047] 4. “Flow field-structure-fatigue” full-chain simulation closed loop: From unsteady aerodynamic simulation to composite stress calculation and fatigue life prediction, a full-chain simulation process is established. It does not rely on bench tests and can complete the high-cycle fatigue performance evaluation of turbine rotors in the design stage, shortening the R&D cycle by 40% and reducing test costs by 60%.

[0048] It should be noted that the above embodiments are merely illustrative examples of the present invention, intended to help understand the technical solution and core ideas of the present invention. Those skilled in the art should understand that any modifications, equivalent substitutions, or improvements made based on the concept of the present invention without departing from its principles should be considered within the scope of protection of the present invention, and the specific scope of protection is determined by the claims.

Claims

1. A simulation method for high-cycle fatigue of turbine rotors based on unidirectional fluid-structure interaction, characterized in that, First, three-dimensional models of the fluid and solid domains are constructed based on the actual turbine model. Second, transient flow field simulation is conducted to obtain turbine aerodynamic loads, and the aerodynamic loads are verified by analyzing their frequency domain characteristics. Third, transient structural simulation analysis is performed. After mapping aerodynamic loads at different times to the turbine solid structure mesh nodes, the transient stress field of the turbine surface nodes is output by iterating the damping through comparison with experimental data. Finally, transient fatigue life simulation is performed to output the maximum principal stress of the turbine key nodes at each time point. Based on the stress-life method, the turbine rotor life is estimated.

2. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 1, characterized in that, The actual turbine model is a simplified 3D simulation model that includes a turbine guide vane and a turbine rotor. Redundant components that have no effect on the flow field inside the turbine are removed. The simplified 3D simulation model is divided into a fluid domain and a solid domain. The fluid domain includes a guide vane flow domain and a rotor flow domain. The guide vane flow domain is set as a stationary domain, and the rotor flow domain is set as a rotating domain. The solid domain includes the turbine rotor structure. An interface is set between the fluid domain and the solid domain.

3. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 2, characterized in that, Transient flow field simulations were conducted to obtain turbine aerodynamic loads. The aerodynamic loads were verified by analyzing their frequency domain characteristics, including: Mesh the fluid domain; The generated mesh was imported into the pressure-based transient solver of the transient flow field simulation software Fluent. The medium, turbulence model, total pressure at the inlet boundary, total temperature at the inlet boundary, static pressure at the outlet boundary, interface between the guide flow domain and the rotor flow domain, MRF rotating coordinate system and rotational speed, and calculation step size were set in the pressure-based transient solver. Extract aerodynamic loads. When there is a clear periodicity in the blade surface pressure, extract the aerodynamic loads on the turbine surface. According to a certain calculation step size, output the coordinates (X,Y,Z) of the center of the turbine surface grid and the static pressure P respectively to obtain sample data at multiple different times. Generate a "number-fluid grid point coordinates-aerodynamic load" file containing a time series within a period from the sample data. Frequency domain analysis was performed on the pressure-time signals from sample data at different times to analyze whether the peak frequency of energy concentration corresponds to the nozzle passage frequency. and harmonics Blade passing frequency and its harmonics Similarly, we will verify whether the aerodynamic load on the turbine surface matches the theoretical formula and simulation results.

4. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 3, characterized in that, When meshing the fluid domain, the mesh on the turbine blade surface is locally refined to ensure that the number of mesh layers at the 0.096mm gap between the blade and the wall is ≥3, the minimum orthogonal mass of the mesh is ≥0.1, and the first boundary layer Y+≈1 on the turbine blade surface.

5. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 3, characterized in that, The transient solver for pressure-based systems uses an ideal gas as the medium and employs the SST k-omega turbulence model. The inlet boundary is set as a pressure inlet with a total pressure of 950 kPa and a total temperature of 62 °C, and the outlet boundary is set as a pressure outlet with a static pressure of 127 kPa. The interface between the guide vane and rotor domains is set as a sliding mesh. The MRF rotating coordinate system method is used to apply a rotational speed of 90,000 rpm around the +Z axis, and the calculation time step is set to 5*10^6. -7 s; The isobaric specific heat capacity of the ideal gas is Cp = 1006.43 J / (kg·K), the thermal conductivity is 0.0242 W / (m·K), and the dynamic viscosity is 1.7894 × 10⁻⁶. -5 kg / (m·s), molecular weight 28.966 kg / kmol.

6. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 5, characterized in that, Transient structural simulation analysis includes: Draw the turbine solid mesh and export the turbine solid surface mesh nodes: Draw the turbine solid mesh through curvature control. The tetrahedral mesh of the aerodynamic load solid domain is refined at the blade tip root. The solid mesh is similar in size to the turbine surface mesh of the fluid domain. Export the coordinates of the solid mesh points. The fluid mesh and solid mesh are mapped one-to-one using the virtual rotation method. That is, the rotating fluid mesh points are mapped one-to-one with the non-rotating solid mesh points by mesh number, and a "number-solid mesh node coordinate" file is generated. A time-series aerodynamic excitation load is generated using bilinear interpolation and the transfer matrix method: The initial aerodynamic load text files "number-fluid grid point coordinates-aerodynamic load" and "number-solid grid node coordinates" are read. Interpolation is performed based on the bilinear interpolation method to generate an aerodynamic excitation load text file "solid grid point coordinates-aerodynamic excitation". The transfer matrix between the aerodynamic load field and the interpolated solid aerodynamic excitation is calculated. The aerodynamic excitation at all subsequent times is calculated using this transfer matrix to generate a time-series aerodynamic excitation load text file "time-solid grid point coordinates-aerodynamic excitation". Set up the ANSYS Transient Structure calculation module. In this calculation module, set the turbine material, constraints and loads, then import the "Time-Solid Mesh Point Coordinates-Aerodynamic Excitation" file and input the aerodynamic excitation load into the calculation module to calculate the transient stress peak. Compare the calculated transient stress peak with the experimental data, and adjust the damping to ensure that the damping falls within a reasonable range; Output turbine stress field history data, i.e., output a file of "turbine surface grid point coordinates-stress-time".

7. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 6, characterized in that, In the ANSYS Transient Structure calculation module, the turbine material is set to 2A12 aluminum alloy; Constraints and load application: The inner cylindrical surface of the turbine is set as a Cylindrical Support to restrict radial and axial displacement, while allowing tangential motion; a rotational speed of 90,000 rpm is applied.

8. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 7, characterized in that, Perform transient fatigue life simulation, including: Read the turbine mesh node coordinates and stress history data from the "Turbine Surface Mesh Point Coordinates-Stress-Time" file; Rainflow calculations were performed on the stress history of each grid point on the turbine surface to obtain the stress cycle amplitude and average stress corresponding to each grid point on the turbine surface. The stress amplitude and average stress of an asymmetric cycle are equivalent to the stress amplitude of a symmetric cycle. By fitting the SN data in the high-cycle region of the turbine, SN curves for different turbine survival rates are constructed; Estimate the lifespan of each grid point on the turbine.

9. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 8, characterized in that, By fitting the SN data of the high-cycle region of the turbine, SN curves for different turbine survival rates are constructed, including: The SN data of the high-frequency region of 2A12 aluminum alloy were fitted using the Basquin formula; Introducing the effect of surface roughness Surface treatment and correction Residual stress correction and size effect factor Correct the SN data; The fatigue life of 2A12 aluminum alloy is assumed to follow a log-normal distribution, and the logarithmic standard deviation of fatigue life is 0.3 under the same stress level. SN curves corresponding to survival rates of 90%, 99%, and 99.9% are constructed.

10. The high-cycle fatigue simulation method for turbine rotors based on unidirectional fluid-structure interaction according to claim 9, characterized in that, The lifespan of each grid point of the turbine is estimated using Miner's linear damage accumulation rule: Let the theoretical fatigue life of the i-th stress cycle at the corresponding stress level be... The damage caused by this cycle is The total damage D is obtained by accumulating the damage from each cycle. Turbine grid point lifespan .

Images