A numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness

Abstract

The present invention discloses a numerical simulation prediction method for the deposition characteristics of thermal barrier coatings taking into account surface roughness, including establishing a grid model, establishing a particle trajectory correction method based on probability density, establishing a deposition model combining a critical velocity model and a particle peeling model, setting a time magnification factor, etc. Based on the statistical correction method, a large number of coating roughness angles are measured to obtain their normal distribution, and then the particle collision trajectory is randomly corrected in real time according to the distribution to simulate the actual collision effect of particles on the smooth model surface. The critical velocity model is used to determine whether the particles meet the preliminary deposition conditions, and the particle peeling model is combined to determine whether the deposition is stable. It has the advantage of high simulation accuracy in coating deposition prediction. The method of the present invention can effectively predict the particle deposition situation on the surface of turbine blades, and provide data support and guidance for the design and timely cleaning of turbine blades.

Description

Technical Field

[0001] The present invention belongs to the technical field, and in particular relates to a numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness. Background Art

[0002] The deposition of thermal barrier coatings (TBCs) on aircraft engine turbine blades is a complex process, significantly impacting the blade's aerodynamic efficiency and heat transfer performance. As critical components, turbine blades are subject to a variety of environmental factors, including extreme operating conditions such as high temperatures and high speeds, which can exacerbate particle deposition and, in turn, impact overall engine efficiency. Therefore, predicting and controlling the deposition characteristics of TBCs on rough surfaces is crucial for maintaining reliable engine operation.

[0003] Numerous studies have shown that the influence of surface roughness on the particle deposition process cannot be ignored. Geometric irregularities on the surface, ranging from micron to submicron levels, can significantly alter key parameters such as particle incident angle distribution, kinetic energy dissipation mechanisms, and adhesion probability. Studies have shown that when the substrate surface roughness Ra exceeds 3.2 μm, the particle contact time is approximately 40% shorter than on smooth surfaces, resulting in an 18% systematic shift in the critical adhesion velocity threshold. This change in the microscopic contact mechanism makes it difficult for traditional simulation models that ignore surface roughness to accurately predict the degradation of the coating's service performance. Summary of the Invention

[0004] The present invention proposes a numerical simulation prediction method for the deposition characteristics of thermal barrier coatings taking into account surface roughness, and proposes a particle deposition prediction method that avoids the establishment of a complex roughness model while ensuring that the prediction relative error is less than 10% (test results show that the maximum error is 7%). A numerical simulation prediction method for the deposition characteristics of thermal barrier coatings taking into account surface roughness is developed to solve the technical problems existing in the background technology.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness, comprising:

[0006] Step 1: Scan the laboratory coating sample with an electron microscope to obtain the coating thickness point cloud at a sampling rate of 5μm and the coating roughness at the same time;

[0007] Step 2: Establish a mesh model: Based on the shape and size of the deposition surface and the mesh size specified in conjunction with the electron microscope sampling rate, use ICEM software to establish a physical model of the deposition surface and divide it into hexahedral meshes to obtain a mesh model of the deposition surface;

[0008] Step 3: Use the deposition surface grid position coordinates to match the coating thickness point cloud coordinate data, and store the coating thickness information in the deposition surface UDM (user-defined memory) at the corresponding position for later use;

[0009] Step 4: Establish a random sweep-based coating roughness angle statistics method: Use the surface sweep macro built into the UDF to scan each grid on the deposition surface. At the same time, determine whether the currently scanned grid is adjacent to the previously scanned grid. If so, call the coating thickness information stored in the two grids to obtain their height difference. Combined with the distance between the two grids, calculate the local roughness angle. By counting all roughness angles that meet the conditions, a normal distribution of the deposition surface roughness angle can be obtained.

[0010] Step 5: Use the obtained normal distribution and the Box-Muller algorithm to generate random particle collision correction angles that conform to this distribution. Considering the classic particle collision "shadow zone" correction method, a correction coefficient is added to the particle collision angle and correction regulations are made. Combined with step 4, a probability density-based particle trajectory correction method is established;

[0011] Step 6: Establish a particle deposition model that couples the critical velocity model with the particle detachment model;

[0012] Step 7: Import the above-mentioned grid model into Ansys-Fluent software, set the numerical simulation boundary conditions, load the particle trajectory correction method based on probability density and the particle deposition model, set the iterative hook function, and input the amplification factor into the hook function;

[0013] Step 8: Perform steady-state calculations on the flow field. Once the flow field is stable and filled with particles, activate the particle collision trajectory correction method and the particle deposition model. After the mesh surface is scanned for the first time and the corresponding normal distribution of roughness angles is obtained, the coating roughness angle statistics method is activated again every hour of deposition to obtain a real-time roughness angle distribution and improve prediction accuracy.

[0014] Step 9: Calculate the particle deposition characteristics of the deposition surface and obtain the local grid deposition layer thickness and deposition rate.

[0015] Furthermore, in step 1, the scanning sampling rate can be adjusted autonomously according to the accuracy of the equipment, and the initial point cloud data obtained needs to be converted into actual coating thickness information.

[0016] Furthermore, in step 2, the mesh model is mesh-independent and meets the minimum mesh quality higher than 0.25.

[0017] Furthermore, in step 3, the method for determining the position matching of the coating thickness point cloud data is:

[0018] ;

[0019] in, is the position coordinate of the model, is the location information contained in the point cloud data, is the device sampling rate, in μm; when the discriminant is true, the coating thickness information is stored in the UDM of the local grid; the coating thickness information is unified and stored in the UDM. At the same time, if the model mesh used in the calculation does not require high precision, its grid size may be larger than the sampling rate. In this case, the above discriminant still applies. It should be noted that this prediction method is still feasible when the scanned coating thickness information is missing. The coating thickness information can be unified and stored in the UDM, and the coating roughness angle distribution is directly assigned to the normal distribution in the prediction method. In this example, the coating thickness is 300 μm, the initial coating roughness is 10.2 μm, the roughness angle distribution has a mean of 0.1271 rad, and a standard deviation of 0.1896 rad.

[0020] Furthermore, in step 4, the core of the coating roughness angle statistics method based on random sweeping is that there are two corresponding grid units on the currently scanned grid surface, namely the c0 unit and the c1 unit. Call the UDF built-in macro to find the grid surface on the c0 or c1 unit. Similarly, find the c0_pre unit and the c1_pre unit of the previous scanned surface, and search for the grid surfaces on the two units. When there is more than one common surface on the grid surfaces of the c0, c1, c0_pre, and c1_pre units, it is judged that the grid units with common surfaces have adjacent grids. The basis for using this method is that when fluent sweeps the target domain, the sweeping mode of a single core is a fan-shaped or spherical development mode that diverges from the starting point to the surrounding areas. In this process, the previous grid unit and the current grid unit have a greater probability of being adjacent grids.

[0021] Furthermore, in step 4, the roughness angle The calculation method is:

[0022] ;

[0023] Among them, atan() is the inverse function of the tangent function, The coating thickness stored for the current scan grid, The coating thickness stored for the last scanned mesh, is the distance between two adjacent grids. When the grid size of the deposition surface remains unchanged, it is directly set to a constant value. Count all the scanned roughness angles and calculate their average value. AVG and standard deviation SD .

[0024] Furthermore, in step 5, the method for generating a random particle collision correction angle that conforms to the normal distribution of the actual collision angle is:

[0025] Generate two independent numbers 、 ; At the same time, the two numbers obey the distribution U(0,1); then generate random values ​​that conform to the standard normal distribution , which is generated as follows:

[0026] ;

[0027] A normal distribution is then generated that fits the corrected angular distribution of the coating:

[0028] .

[0029] Furthermore, in step 5, the consideration of the “shadow area” is to add a correction factor :

[0030] When the roughness angle is negative and its absolute value is greater than the particle incident angle, the area is considered to be a shadow area, and it is believed that particles will not collide in this area, and its probability is zero;

[0031] ;

[0032] When the roughness angle is negative, but its absolute value is greater than zero and smaller than the particle incident angle, the probability of the roughness angle being negative is considered to be smaller than the probability of hitting the horizontal wall, and a correction factor is added;

[0033] ;

[0034] The probability of a positive roughness angle is higher than the probability of a zero roughness angle, so the correction factor is:

[0035] ;

[0036] The obtained correction factor is multiplied by the correction angle generated by the Box-Muller algorithm:

[0037] ;

[0038] in, The function is the roughness angle correction factor output by the program. This factor further corrects the coating roughness angle distribution to a real collision distribution that takes into account the "shadow area" effect. The final particle collision correction angle, is the particle incident angle, is the coating roughness angle, is a negative roughness angle, A positive roughness angle.

[0039] Furthermore, in step 6, the deposition model of the critical velocity model and the particle exfoliation model is:

[0040] First, the critical velocity model is used to determine whether the colliding particles meet the deposition conditions. The critical velocity model studies the elastic properties of the particles and uses the normal velocity of the particles when they collide with the wall as the basis for judgment. When the normal velocity of the particles is greater than the critical velocity, the particles will rebound and no deposition will occur. When the normal velocity of the particles is less than the critical velocity, the particles will adhere to the wall.

[0041] ;

[0042] ;

[0043] ;

[0044] is the critical speed, is the Young's modulus of the composite material, is the surface Young's modulus, is the Young's modulus of the particle and satisfies the temperature , is the surface Poisson's ratio, is the particle Poisson's ratio, is the particle radius (consistent with the critical velocity model parameter), is the particle density, is the process quantity;

[0045] The particle detachment model is then used to determine whether the particles can be stably deposited. The reason for using additional judgment is that when the wall detachment shear velocity experienced by the adhered particles in the boundary layer is greater than the critical wall detachment shear velocity, the adhered particles will detach from the wall and rejoin the mainstream:

[0046] ;

[0047] ;

[0048] is the critical wall shear velocity of the fluid, is the Cunningham correction factor, is the particle adhesion work, measured by experiment, is the fluid density, is the composite Young's modulus;

[0049] in, Described as:

[0050] ;

[0051] ;

[0052] is the Knudsen number, which indicates the relative size of the free path of gas molecules and the particle size. is the average air travel, described as a diameter of The average distance between two adjacent collisions of a spherical molecule is the mean free path of the molecule, which is expressed as:

[0053] ;

[0054] is the effective molecular diameter, which is 3.5 10 -10 m, is the number of molecules per unit volume, which can be calculated using the ideal gas law as:

[0055] ;

[0056] is the pressure, the Boltzmann constant 1.38 10 -23 J / K, is the gas temperature in K;

[0057] The shear rate received by the adhering particles is:

[0058] ;

[0059] is the wall shear stress.

[0060] Furthermore, in step 7, the iterative hook function specifies the service time represented by each iteration. , and add the time magnification factor , as the Ansys-Fluent software is gradually iterated, the service time gradually accumulates, where the service time is described as:

[0061] ;

[0062] Where, For the length of service, is the length of service, is the time magnification factor, is the number of iterations, is the service time represented by a single iteration.

[0063] Furthermore, in step 8, the time interval for reactivating the coating roughness angle statistics method can be actively adjusted as needed, with the highest frequency being one sweep activated for every one iteration.

[0064] Furthermore, in step 9, the deposition rate The expression is:

[0065] ;

[0066] In the formula is the deposition rate, is the cumulative deposition mass of particles, is the current grid area, Length of service;

[0067] Deposition thickness The expression is:

[0068] ;

[0069] Where, is the particle density.

[0070] The numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness of the present invention has the following advantages:

[0071] (1) The prediction method of the present invention rapidly obtains thermal barrier coating deposition characteristic data through numerical simulation. Utilizing the user-defined function (UDF) in Ansys-Fluent and considering the basic order of the computational core traversing the grid cells, a statistical method for the random sweep coating roughness angle that rapidly identifies adjacent grids is developed. Based on this statistical method, a normal distribution of the coating roughness angle is constructed. The obtained roughness angle distribution is then connected with the Box-Muller algorithm. Combined with the classic study of the influence of the coating "shadow area" on the particle collision angle, a particle trajectory correction angle that conforms to the actual particle collision effect is successfully output, achieving the simulation of the particle deposition characteristics of a rough wall on a smooth wall. The method of the present invention is only applied in steady-state flow fields. The simulation value before stabilization will not affect the calculation results. At the same time, the application of this method avoids the tedious construction of small-scale coating physical models and meshing, has low requirements on computing resources, improves computing efficiency, and better meets the needs of actual engineering applications.

[0072] (2) The prediction method of the present invention is based on the classical particle deposition model. The critical velocity model is combined with the particle peeling model to determine whether the collision particles are stably deposited. The deposition process of the coating surface is numerically simulated using Fluent software. At the same time, the deposition results are compared and verified with the real rough model to ensure the accuracy of the model. The particle deposition distribution on the coating surface at any time in the future is obtained, and the deposition characteristics of the thermal barrier coating are predicted.

[0073] (3) By avoiding the need to establish a real physical model of the coating, the present invention simplifies the calculation process and saves computing resources. At the same time, the concept of amplification factor is introduced, which greatly reduces the time required for numerical simulation and achieves fast and accurate coating deposition prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A flowchart for implementing a numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness according to the present invention;

[0075] Figure 2 This is the mesh division result diagram of the present invention;

[0076] Figure 3 Schematic diagram of the particle collision angle and coating roughness angle of the present invention;

[0077] Figure 4 A flow chart for implementing the particle trajectory correction method based on probability density of the present invention;

[0078] Figure 5 A statistical roughness angle distribution diagram for the program of the present invention;

[0079] Figure 6 is the actual roughness angle distribution of the coating of the present invention;

[0080] Figure 7 is a deposition thickness distribution diagram in the embodiment;

[0081] Figure 8 Graph showing the deposition rate distribution in the embodiment. DETAILED DESCRIPTION

[0082] The present invention will be further explained below with reference to the accompanying drawings.

[0083] like Figure 1 As shown, the present invention provides a numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness, which specifically includes the following steps:

[0084] Step 1: Scan the laboratory coating sample with an electron microscope to obtain the coating thickness point cloud at a sampling rate of 5μm and the coating roughness at the same time.

[0085] Step 2: Establish a mesh model: According to the shape and size of the deposition surface and in combination with the electron microscope sampling rate, the mesh size is specified. Then, the physical model of the deposition surface is established using ICEM software and divided into hexahedral meshes to obtain the mesh model of the deposition surface.

[0086] Step 3: Use the deposition surface grid position coordinates to match the coating thickness point cloud coordinate data, and store the coating thickness information in the deposition surface UDM at the corresponding position for later use.

[0087] Step 4: Establish a random sweep-based coating roughness angle statistics method: Use the surface sweep macro built into the UDF to scan each grid on the deposition surface. Simultaneously, determine whether the currently scanned grid is adjacent to the previously scanned grid. If so, retrieve the coating thickness information stored in the two grids to determine their height difference. Combined with the distance between the two grids, the local roughness angle is calculated. By counting all roughness angles that meet the criteria, a normal distribution of the deposition surface roughness angle is obtained.

[0088] Step 5: Use the obtained normal distribution and the Box-Muller algorithm to generate random particle collision correction angles that conform to this distribution. Considering the classic particle collision "shadow zone" correction method, a correction coefficient is added to the particle collision angle and a correction rule is made. Combined with step 4, a probability density-based particle trajectory correction method is established.

[0089] Step 6: Establish a particle deposition model that couples the critical velocity model with the particle detachment model.

[0090] Step 7: Import the above-mentioned grid model into Ansys-Fluent software, set the numerical simulation boundary conditions, load the probability density-based particle trajectory correction method and particle deposition model, set the iterative hook function, and input the amplification factor into the hook function.

[0091] Step 8: Perform steady-state calculations on the flow field. After the flow field stabilizes and is filled with particles, activate the particle collision trajectory correction method and the particle deposition model. After the mesh surface is scanned for the first time and the corresponding normal distribution of roughness angles is obtained, it is stipulated that the coating roughness angle statistics method is activated again every hour of deposition to obtain real-time roughness angle distribution and improve prediction accuracy.

[0092] Step 9: Calculate the particle deposition characteristics of the deposition surface and obtain the local grid deposition layer thickness and deposition rate.

[0093] Among them, the coating deposition characteristics refer to the changes in the coating surface morphology distribution caused by particle deposition, including the coating deposition rate, coating deposition quality and coating deposition thickness. Step 8, while activating the particle trajectory correction method, will also support the activation of the particle deposition model, thereby preparing for the subsequent step 9 to obtain the coating surface deposition morphology.

[0094] After completing the above steps, the entire particle deposition method will be simulated and compared to verify the accuracy of its model. The verified numerical simulation prediction method of the thermal barrier coating deposition characteristics considering surface roughness will be put into practical application to realize the prediction of the deposition status of the turbine blade surface coating at any time.

[0095] In step 1, the roughness characteristics of the target coating need to be collected and the electron microscope sampling rate needs to be determined.

[0096] In step 2, the mesh size of the deposition surface needs to match the sampling rate of the coating roughness feature. If high reproduction accuracy is not required, the point cloud data can be appropriately sparsely processed. At the same time, the mesh must meet the requirements: the minimum quality mesh must be higher than 0.25, that is, the Jacobian value of each hexahedron in the hexahedral mesh must be greater than 0.25. The mesh division result is as follows Figure 2 shown.

[0097] In step 3, the method for determining the position matching of the coating thickness point cloud data is:

[0098] ;

[0099] in, is the position coordinate of the model, is the location information contained in the point cloud data, is the device sampling rate, in μm; when the discriminant is true, the coating thickness information is stored in the UDM of the local grid; the coating thickness information is unified and stored in the UDM. At the same time, if the model mesh used in the calculation does not require high precision, its grid size may be larger than the sampling rate. In this case, the above discriminant still applies. It should be noted that this prediction method is still feasible when the scanned coating thickness information is missing. The coating thickness information can be unified and stored in the UDM, and the coating roughness angle distribution is directly assigned to the normal distribution in the prediction method. In this example, the coating thickness is 300 μm, the initial coating roughness is 10.2 μm, the roughness angle distribution has a mean of 0.1271 rad, and a standard deviation of 0.1896 rad.

[0100] In step 4, the core of the statistical method of coating roughness angle based on random sweeping is that there are two corresponding grid units on the currently scanned grid surface, namely c0 unit and c1 unit. Call the UDF built-in macro to find the grid surface on the c0 or c1 unit. Similarly, find the c0_pre unit and c1_pre unit of the previous scanned surface, and search for the grid surfaces on the two units. When there is more than one common surface on the grid surfaces of the c0, c1, c0_pre, and c1_pre units, it is judged that the grid units with common surfaces have adjacent grids. The basis for using this method is: when fluent sweeps the target domain, the sweeping mode of a single core is a fan-shaped or spherical development mode that diverges from the starting point to the surrounding areas. In this process, the previous grid unit and the current grid unit have a greater probability of being adjacent grids. The specific definitions and distinctions of collision angles and roughness angles are as follows: Figure 3 shown.

[0101] In step 4, rough corners The calculation method is:

[0102] ;

[0103] in, The coating thickness stored for the current scan grid, The coating thickness stored for the last scanned mesh, is the distance between two adjacent grids. When the grid size of the deposition surface remains unchanged, it is directly set to a constant value. Count all the scanned roughness angles and calculate their average value. AVG and standard deviation SD It is important to note the differences between the coating roughness angle, particle collision angle, and particle incidence angle. The coating roughness angle refers to the angle between two adjacent sampling points due to the height difference. The particle collision angle refers to the angle when the particle collides with the rough coating. The particle incidence angle refers to the angle at which the particle enters from the inlet.

[0104] In step 5, the random correction angle method that conforms to the normal distribution of the actual particle collision angle, namely the Box-Muller algorithm, is generated as follows:

[0105] Generate two independent numbers 、 ; At the same time, the two numbers obey the distribution U(0,1); then generate random values ​​that conform to the standard normal distribution , which is generated as follows:

[0106] ;

[0107] A normal distribution is then generated that fits the corrected angular distribution of the coating:

[0108] .

[0109] The specific output process can be obtained by Figure 4 express.

[0110] The considerations for the “shaded area” are:

[0111] When the roughness angle is negative and its absolute value is greater than the particle incident angle, the area is considered to be a shadow area, and it is believed that particles will not collide in this area, and its probability is zero;

[0112] ;

[0113] When the roughness angle is negative, but its absolute value is greater than zero and smaller than the particle incident angle, the probability of the roughness angle being negative is considered to be smaller than the probability of hitting the horizontal wall, and a correction factor is added;

[0114] ;

[0115] The probability of a positive roughness angle is higher than the probability of a zero roughness angle, so the correction factor is:

[0116] ;

[0117] The obtained correction factor is multiplied by the correction angle generated by the Box-Muller algorithm:

[0118] ;

[0119] in, The final particle collision correction angle, is the particle incident angle, is the coating roughness angle, is a negative roughness angle, A positive roughness angle.

[0120] In step 6, the deposition model of the critical velocity model and the particle separation model is:

[0121] First, the critical velocity model is used to determine whether the colliding particles meet the deposition conditions. The critical velocity model studies the elastic properties of the particles and uses the normal velocity of the particles when they collide with the wall as the basis for judgment. When the normal velocity of the particles is greater than the critical velocity, the particles will rebound and no deposition will occur. When the normal velocity of the particles is less than the critical velocity, the particles will adhere to the wall.

[0122] ;

[0123] ;

[0124] ;

[0125] is the critical speed, is the Young's modulus of the composite material, is the surface Young's modulus, is the Young's modulus of the particle and satisfies the temperature , is the surface Poisson's ratio, is the particle Poisson's ratio, is the particle radius (consistent with the critical velocity model parameter), is the particle density, is the process quantity;

[0126] The particle detachment model is then used to determine whether the particles can be stably deposited. The reason for using additional judgment is that when the wall detachment shear velocity experienced by the adhered particles in the boundary layer is greater than the critical wall detachment shear velocity, the adhered particles will detach from the wall and rejoin the mainstream:

[0127] ;

[0128] ;

[0129] is the critical wall shear velocity of the fluid, is the Cunningham correction factor, is the particle adhesion work, measured by experiment, is the fluid density, is the composite Young's modulus;

[0130] in, Described as:

[0131] ;

[0132] ;

[0133] is the Knudsen number, which indicates the relative size of the free path of gas molecules and the particle size. is the average air travel, described as a diameter of The average distance between two adjacent collisions of a spherical molecule is the mean free path of the molecule, which is expressed as:

[0134] ;

[0135] is the effective molecular diameter, which is 3.5 10 -10 m, is the number of molecules per unit volume, which can be calculated using the ideal gas law as:

[0136] ;

[0137] is the pressure, the Boltzmann constant 1.38 10 -23 J / K, is the gas temperature in K;

[0138] In this example, W A The particle adhesion work is a parameter of silicon particles, which is 0.039 J / m 3 .

[0139] The shear rate received by the adhering particles is:

[0140] ;

[0141] is the wall shear stress.

[0142] In step 7, the iteration hook function specifies the service time represented by each iteration , and add the time magnification factor , as the Ansys-Fluent software is gradually iterated, the service time gradually accumulates, where the service time is described as:

[0143] ;

[0144] Where, For the length of service, is the length of service, is the time magnification factor, is the number of iterations, is the service time represented by a single iteration.

[0145] In step 8, the first program outputs the corrected angular distribution as Figure 5 As shown, the left vertical axis and the orange bar graph are the 100 random angle frequency distributions output by the program, while the right vertical axis and Figure 5 The red curve in the middle refers to the normal distribution fitting result of the frequency distribution. Figure 5 The horizontal axis in the figure refers to the correction angle output by the program, and the coating roughness angle distribution swept by the program is as follows: Figure 6 As shown, the vertical axis is the probability density of the rough angle swept by the program, and the horizontal axis is the size of the rough angle. The black curve is the actual probability density distribution of the coating rough angle, and the red curve refers to the normal distribution of the coating rough angle obtained by fitting after the program sweeps the user-defined memory. It is found that there is an error between the two, which can be reduced by the "shadow area" correction coefficient. The time interval for reactivating the coating rough angle statistical method can be actively adjusted as needed, and the highest frequency is to activate a sweep once per iteration.

[0146] In step 9, the deposition rate The expression is:

[0147] ;

[0148] In the formula is the deposition rate, is the cumulative deposition mass of particles, is the current grid area, Length of service;

[0149] Deposition thickness The expression is:

[0150] ;

[0151] Where, is the particle density. The coating surface deposition thickness distribution results are as follows Figure 7 The deposition rate distribution is shown in Figure 8 shown.

[0152] The present invention provides an embodiment specifically comprising:

[0153] (1) Obtain laboratory coating point cloud data, the resolution is determined by the equipment.

[0154] (2) Establish a grid model.

[0155] Based on the shape and size of the coating blade, a microscopic three-dimensional geometric model of the particle deposition flow field is established. According to the actual situation in the calculation flow field, the ICEM software is used to divide the structured grid into hexahedral grids to achieve discretization of the calculation domain.

[0156] To accurately simulate particle deposition on the coating surface, the deposition surface mesh size must be matched to the coating data point cloud sampling rate. A boundary layer mesh is set at the fluid-structure interaction interface in the computational domain. The specific size is calculated using Y+. Periodic boundaries are then set above and below the flow channel to simulate deposition in the cascade channel. Finally, the mesh parameters are checked and adjusted to ensure that the overall model mesh quality reaches above 0.25.

[0157] (3) Matching the coating data point cloud with the mesh model.

[0158] Due to the lack of point cloud data for typical blade surface coatings, the coating thickness was uniformly set to 300 μm. At the same time, the known coating roughness characteristics, namely the normal distribution of roughness angles, were directly assigned to the first sweep result.

[0159] (4) A particle trajectory correction method based on probability density was established. The sequential logic of the grid cells was used in Ansys-Fluent software to establish a statistical method for the roughness angle of the coating based on random sweeping. The method judged whether the grid surfaces of two adjacent scans were adjacent grids. If the result was true, the roughness angle was calculated.

[0160] (5) Establish the normal distribution of coating roughness angle, and count all roughness angles that meet the judgment conditions to establish the normal distribution of the roughness angle of the target coating. At the same time, the classic collision "shadow zone" effect is added to build a complete coating roughness angle output logic.

[0161] (6) Add a particle deposition model that couples the critical velocity model with the particle stripping model, and compile the above functional UDF and load it.

[0162] (7) Set boundary conditions, open energy, discrete phase model, set the iteration hook function to link the number of iteration steps to the service time, and perform deposition calculation. The deposition rate (g, deposition mass per unit area per unit time) within a certain service time and the coating surface deposition thickness data are stored in the user-defined memory UDM and exported. The roughness angle normal distribution constructed by program sweeping is compared with the actual coating roughness angle distribution. The comparison results are as follows: Figure 5 、 Figure 6 As shown in Figure 2, the error between the roughness angle distribution of the predicted results and the actual model is within the acceptable range, which proves the accuracy of the model.

[0163] (8) Perform steady-state calculations on the flow field and update the roughness angle distribution.

[0164] (9) Example Result Analysis: This example studies the particle deposition characteristics of a coating applied on the surface of a typical turbine blade. The predicted distribution of particle deposition on the blade surface during a certain period of service is as follows: Figure 7 This demonstrates that the proposed prediction method can simulate the deposition of coatings on blade surfaces over any service life. This method saves significant computational resources and time while maintaining high simulation accuracy, and can predict changes in characteristic parameters such as the thickness and position of the deposited layer on turbine blades over their service life.

[0165] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A numerical simulation prediction method for thermal barrier coating deposition characteristics considering surface roughness, characterized in that: include: Step 1: Scan the coating sample with an electron microscope to obtain coating thickness point cloud data at a preset sampling rate, and also obtain coating roughness; Step 2: According to the shape and size of the deposition surface and the sampling rate of the electron microscope, the grid size is specified. Then, the physical model of the deposition surface is established using ICEM software and divided into hexahedral grids to obtain the grid model of the deposition surface. Step 3: Use the deposition surface grid position coordinates to match the coating thickness point cloud coordinate data, and store the coating thickness information in the deposition surface UDM at the corresponding position for later use; Step 4: Establish a statistical method for coating roughness angle based on random sweep: Use the surface sweep macro built into the UDF to scan each grid divided on the deposition surface, and at the same time determine whether the scanned grid is an adjacent grid of the last scanned grid. If so, call the coating thickness information stored in the two grids to obtain their height difference, and calculate the local roughness angle based on the distance between the two grids. ; Count all rough angles that meet the conditions , we can obtain the normal distribution of the roughness angle of the deposition surface; Step 5: Call the obtained normal distribution and use the Box-Muller algorithm to generate random particle collision correction angles that conform to the distribution; consider the classic particle collision shadow area correction method, add a correction coefficient on the basis of the particle collision angle and make correction regulations, and combine step 4 to establish a probability density-based particle trajectory correction method; Step 6: Establish a particle deposition model that couples the critical velocity model with the particle detachment model; Step 7: Import the mesh model into Ansys-Fluent software, set the numerical simulation boundary conditions, load the probability density-based particle trajectory correction method and particle deposition model, set the iterative hook function, and input the amplification factor into the hook function; Step 8: Perform steady-state calculations on the flow field. Once the flow field is stable and filled with particles, activate the particle collision trajectory correction method and the particle deposition model. After the mesh surface is scanned for the first time and the corresponding normal distribution of roughness angles is obtained, the coating roughness angle statistics method is activated again every several hours of deposition to obtain a real-time roughness angle distribution. Step 9: Calculate the particle deposition characteristics of the deposition surface and obtain the local grid deposition layer thickness and deposition rate.

2. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 2, the mesh model is verified to be mesh-independent and to meet the minimum mesh quality higher than 0.

25.

3. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 3, the method for determining the position matching of the coating thickness point cloud data is: ； in, is the position coordinate of the model, is the location information contained in the point cloud data, is the device sampling rate, in μm; when the discriminant is true, the coating thickness information is stored in the UDM of the local grid; the coating thickness information is unified and stored in the UDM, and the coating roughness angle distribution is directly assigned to the normal distribution in the prediction method.

4. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 4, the coating roughness angle statistics method based on random sweeping specifically includes the following steps: there are two corresponding grid units on the currently scanned grid surface, namely, the c0 unit and the c1 unit; calling the UDF built-in macro to find the grid surface on the c0 or c1 unit; finding the c0_pre unit and the c1_pre unit of the previous scanned surface, and searching for the grid surfaces on the two units; when there is more than one common surface on the grid surfaces of the c0, c1, c0_pre, and c1_pre units, it is determined that the grid units with the common surface have adjacent grids.

5. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 4, the rough angle The calculation method is: ； Among them, atan() is the inverse function of the tangent function, The coating thickness stored for the current scan grid, The coating thickness stored for the last scanned mesh, is the distance between two adjacent grids. When the grid size of the deposition surface remains unchanged, it is directly set to a constant value. Count all the scanned roughness angles and calculate their average value. AVG and standard deviation SD .

6. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 5, characterized in that: In step 5, the method for generating a random particle collision correction angle that conforms to the normal distribution of the actual collision angle is: Generate two independent numbers 、 ; At the same time, the two numbers obey the distribution U(0,1); then generate random values ​​that conform to the standard normal distribution , which is generated as follows: ； A normal distribution is then generated that fits the corrected angular distribution of the coating: 。 7. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 6, characterized in that: In step 5, the shadow area is considered by adding a correction factor : When the roughness angle is negative and its absolute value is greater than the particle incident angle, the area is considered to be a shadow area, and it is believed that particles will not collide in this area, and its probability is zero; ； When the roughness angle is negative, but its absolute value is greater than zero and smaller than the particle incident angle, the probability of the roughness angle being negative is considered to be smaller than the probability of hitting the horizontal wall, and a correction factor is added; ； The probability of a positive roughness angle is higher than the probability of a zero roughness angle, so the correction factor is: ； The obtained correction factor is multiplied by the correction angle generated by the Box-Muller algorithm: ； in, The function is the rough angle correction factor output by the program, which corrects the coating rough angle distribution to the real collision distribution taking into account the shadow area effect. The final particle collision correction angle, is the particle incident angle, is the coating roughness angle, is a negative roughness angle, A positive roughness angle.

8. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 6, the deposition model of the critical velocity model and the particle exfoliation model is: First, the critical velocity model is used to determine whether the colliding particles meet the deposition conditions. The critical velocity model studies the elastic properties of the particles and uses the normal velocity of the particles when they collide with the wall as the basis for judgment. When the normal velocity of the particles is greater than the critical velocity, the particles will rebound and no deposition will occur. When the normal velocity of the particles is less than the critical velocity, the particles will adhere to the wall. ； ； ； is the critical speed, is the Young's modulus of the composite material, is the surface Young's modulus, is the Young's modulus of the particle and satisfies the temperature , is the surface Poisson's ratio, is the particle Poisson's ratio, is the particle radius, is the particle density, is the process quantity; The particle detachment model is then used to determine whether the particles can be stably deposited. The reason for using additional judgment is that when the wall detachment shear velocity experienced by the adhered particles in the boundary layer is greater than the critical wall detachment shear velocity, the adhered particles will detach from the wall and rejoin the mainstream: ； ； is the critical wall shear velocity of the fluid, is the Cunningham correction factor, is the particle adhesion work, measured by experiment, is the fluid density, is the composite Young's modulus; in, Described as: ； ； is the Knudsen number, which indicates the relative size of the free path of gas molecules and the particle size. is the average air travel, described as a diameter of The average distance between two adjacent collisions of a spherical molecule is the mean free path of the molecule, which is expressed as: ； is the effective molecular diameter, which is 3.5 10 -10 m, is the number of molecules per unit volume, which can be calculated using the ideal gas law as: ； is the pressure, the Boltzmann constant 1.38 10 -23 J / K, is the gas temperature in K; The shear rate received by the adhering particles is: ； is the wall shear stress.

9. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 7, the iteration hook function specifies the service time represented by each iteration. , and add the time magnification factor , as the Ansys-Fluent software is gradually iterated, the service time gradually accumulates, where the service time is described as: ； Where, For the length of service, is the length of service, is the time magnification factor, is the number of iterations, is the service time represented by a single iteration.

10. The method for numerical simulation prediction of thermal barrier coating deposition characteristics considering surface roughness according to claim 1, characterized in that: In step 9, the deposition rate The expression is: ； In the formula is the deposition rate, is the cumulative deposition mass of particles, is the current grid area, Length of service; Deposition thickness The expression is: ； Where, is the particle density.

Images