Numerical Simulation Method and Apparatus for Supersonic Liquid Jet Atomization

Abstract

This invention proposes a numerical simulation method and apparatus for supersonic liquid jet atomization, aiming to address the numerical instability and jet-grid scale dependence issues of the traditional Eulerian-Lagrange method in scenarios simulating droplets larger than the grid. The method includes: proposing an improved pre-stored neighbor list strategy for data preprocessing; solving the gas-phase Navier-Stokes equations using the liquid droplet size and a large array of pre-stored near-orifice region grid neighbors; freezing the gas-phase Navier-Stokes equations and solving the liquid-phase governing equations using the Runge-Kutta method; calculating and utilizing the source terms on each grid to correct the flow field given by the frozen gas-phase Navier-Stokes equations, thus updating the basic flow field physical quantities. This invention saves computational resources by moving a large amount of neighbor grid searching to the preprocessing stage, achieving stable simulation of jets of arbitrary sizes, and has broad engineering application prospects.

Description

Technical Field

[0001] This invention belongs to the field of fluid dynamics calculation and data processing technology, and in particular, it is a numerical simulation method and device for supersonic liquid jet atomization. Background Technology

[0002] In the field of supersonic liquid jet atomization technology, the Eulerian-Lagrange method is one of the most common numerical simulation tools for jet atomization. However, traditional Eulerian-Lagrange methods have significant limitations when performing numerical simulations of jet atomization: numerical instability with large droplets and jet-mesh scale dependence. For example, when the droplet size is larger than the mesh resolution, significant calculation errors occur in the results of droplet breakup, trajectory, and gas flow field, and calculation divergence is very likely to occur, failing to accurately reflect the actual physical process. The finer the mesh, the more obvious these limitations become. Finally, GPUs and finite volume methods have become the mainstream means of numerical simulation in industry. Performing a large number of neighbor mesh searches under this framework also faces challenges and low computational efficiency. Furthermore, implementing improvements to the Eulerian-Lagrange method requires a large amount of neighbor mesh search work. All of these factors greatly limit the application of the model in complex supersonic liquid jet atomization scenarios, making it difficult to meet the current demand for accurate simulation and analysis of this process to improve related industrial design and performance optimization. Summary of the Invention

[0003] This invention addresses the numerical instability and jet-mesh scale dependence of traditional Eulerian-Lagrange methods in scenarios where the simulated droplet size is larger than the mesh. It proposes a numerical simulation method and apparatus for supersonic liquid jet atomization. By placing a significant amount of neighbor mesh searching during preprocessing, it substantially reduces the computational resources required for neighbor mesh searching within the framework of GPUs and finite volume methods. This enables stable simulation of liquid jets of arbitrary size with reasonable simulation results. Simultaneously, it eliminates the dependence of the numerical jet size on the mesh, reducing empirical errors caused by manually setting jet parameters.

[0004] A numerical simulation method for supersonic liquid jet atomization includes the following steps:

[0005] Step 110: Preprocess the gas-phase Navier-Stokes equations for supersonic liquid jet atomization, including using an improved pre-stored neighbor list strategy to establish a large array to pre-store the grid neighbors of the near-orifice region.

[0006] Step 120: Using the droplet size of the liquid and the pre-stored near-nozzle region grid neighbors in the large array, solve the gas phase Navier-Stokes equations to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature;

[0007] Step 130: Freeze the gas phase Navier-Stokes equations, and obtain information about the grid around the droplet by using the relationship between the droplet diameter and the grid size where the droplet is located; substitute the information about the grid around the droplet into the liquid phase control equations, and solve the liquid phase control equations using the Runge-Kutta method;

[0008] Step 140: Calculate the source terms carried on each grid.

[0009] Step 150: Use the source terms on each grid to correct the flow field given by the frozen gas-phase Navier-Stokes equations and update the basic flow field physical quantities for the next time step.

[0010] Step 160: Repeat steps 120 to 150 until the updated basic flow field physical quantities at all times are obtained.

[0011] On the other hand, the present invention provides a numerical simulation device for supersonic liquid jet atomization, which implements the steps of the aforementioned numerical simulation method for supersonic liquid jet atomization. The device includes:

[0012] The first module is used to preprocess the gas-phase Navier-Stokes equations for supersonic liquid jet atomization, including using an improved pre-stored neighbor list strategy to establish a large array to pre-store the grid neighbors of the near-orifice region.

[0013] The second module is used to solve the gas phase Navier-Stokes equations using the liquid droplet size and the pre-stored near-nozzle region grid neighbors in a large array, to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature;

[0014] The third module is used to freeze the Navier-Stokes equations in the gas phase. It uses the relationship between the droplet diameter and the grid size where the droplet is located to obtain information about the grid around the droplet. The information about the grid around the droplet is substituted into the liquid phase control equations, and the Runge-Kutta method is used to solve the liquid phase control equations.

[0015] The fourth module is used to calculate the source terms carried on each grid.

[0016] The fifth module is used to correct the flow field given by the frozen gas-phase Navier-Stokes equations using source terms on each grid, and update the basic flow field physical quantities for the next time step.

[0017] The sixth module is used to repeatedly execute the steps of the second to fifth modules until the updated basic flow field physical quantities at all times are obtained.

[0018] Compared with the prior art, the technical effects of the present invention include:

[0019] (1) By proposing an improved pre-stored neighbor list strategy, we can specifically address the large amount of neighbor grid search work required by the finite volume method under the GPU architecture. Through centralized computation in the preprocessing stage, we can significantly reduce the retrieval burden in the real-time computation stage and save a lot of computing resources.

[0020] (2) In view of the characteristics of the multi-droplet system in the spray field, the source term allocation scheme of the nearest node kernel function combined with the Gaussian kernel function is adopted. The source term is first allocated to the nearest grid node and then diffused to the surrounding grid within a certain range, thus avoiding the instability of large droplet injection simulation caused by excessive concentration of source terms.

[0021] (3) By introducing the upstream averaging method, the problem of underestimation of penetration depth caused by droplet self-disturbance in the classical method is solved. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the steps of a numerical simulation method for supersonic liquid jet atomization in one embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram illustrating the implementation process of the pre-stored neighbor table algorithm in one embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of a method for obtaining information about the grid around a droplet in one embodiment of the present invention, wherein, Figure 3 (a) is a schematic diagram showing the acquisition of information about the grid around the droplet using the upstream averaging method. express The area within, It is the standard deviation of the Gaussian function. It is the droplet diameter. It is the droplet radius. Figure 3 (b) is a schematic diagram illustrating the acquisition of information about the grid around the droplet using the inverse distance weighting method;

[0026] Figure 4 This is a schematic diagram illustrating the source term allocation of the nearest node kernel function and the Gaussian kernel function in one embodiment of the present invention, wherein, Figure 4 (a) is a schematic diagram of source term allocation given by the nearest node kernel function. Figure 4(b) is a schematic diagram of the source term allocation given by the Gaussian kernel function;

[0027] Figure 5 This is a flowchart illustrating the entire calculation process of a numerical simulation of supersonic liquid jet atomization in one embodiment of the present invention.

[0028] Figure 6 This is a schematic diagram of the simulation domain, mesh, and discrete droplet jetting in the experiment of this invention;

[0029] Figure 7 This is a schematic diagram comparing the quantitative results of penetration depth obtained using different bidirectional coupling methods in the experiments of this invention with experimental results. Indicates the depth of penetration. The initial injection diameter is shown on the ordinate. , For space Coordinate position;

[0030] Figure 8 This is a schematic diagram of the droplet distribution results in a grouped calculation example using the method of this invention in the experiment of this invention, wherein, Figure 8 (a) Selecting examples for group b1 The droplet distribution results at that time Figure 8 (b) Selecting examples for group b2 The droplet distribution results at that time Figure 8 (c) Selecting examples for group b3 The droplet distribution results at that time. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] In one embodiment, such as Figure 1 As shown, this invention provides a numerical simulation method for supersonic liquid jet atomization, comprising:

[0033] Step 110: First, the gas-phase Navier-Stokes (NS) equations to be solved are preprocessed, and then an improved pre-stored neighbor list strategy is adopted to create a large array to pre-store the grid neighbors of the near-orifice region for subsequent calculations.

[0034] Specifically, the Navier-Stokes equations with droplet source terms are as follows:

[0035] (1)

[0036] Where, on the left side of the equation: For the incoming Reynolds number, These are the conserved quantities in the flow field (including mass conservation, momentum conservation, and kinetic energy conservation). For gas phase density, It is the gas phase velocity vector. It is gas phase energy; , , Coordinate systems Middle Inviscid flux in the direction, , , for Viscous flux in the directional direction (see: Li Peibo. Numerical simulation of mixing and combustion processes of transverse spray in supersonic airflow [D]. Changsha: National University of Defense Technology, 2019: 25). Right side of the equation: It is the column vector of the droplet source terms in the gas-phase Navier-Stokes equations, and the source terms... , It is a quality source item. It is a momentum source term. It is an energy source term. This indicates the transpose of a vector or matrix.

[0037] During the supersonic liquid jet atomization process, using Represents the basic flow field physical quantities. It is the gas phase density. These are the gas phase velocities in three coordinate directions. It is gas phase pressure. It refers to the gas phase temperature. Specifically, step 110 includes:

[0038] Step 111 involves initializing the variables in the gas phase NS equations, performing mesh coordinate transformation, computation array allocation, and mesh preprocessing.

[0039] Step 112, Define the coordinate system Coordinates of the near-middle nozzle area The range, as the near-orifice area An improved pre-stored neighbor list strategy is adopted to filter out those falling into the near-orifice region. The grid index is recorded in the grid list of the near-jet port area. The specific algorithm is shown in Algorithm 1, which is described in the pseudocode in Table 1.

[0040] Table 1: Pseudocode for the filtering network index algorithm in the near-orifice region

[0041]

[0042] Here, the index "Tid" represents the grid index in the global computation domain. This represents the location vector of the grid (global grid) within the global computational domain. This represents the total number of grids in the global computational domain. In step 3, the function "atomicAdd" is an atomic operation in GPU parallel computing. It adds the specified value to a memory location, ensuring that the addition operation can be executed as a single, uninterrupted operation even in a multi-threaded or concurrent environment. The "gridCounter" indicates the total number of grids in the near-injection region, which is a global variable used as input. "1" indicates that the increment step is 1 each time. The integer variable "index" allocates a contiguous storage location for each grid in the near-injection region. This includes storing the original grid index in step 4 through the near-injection region grid list array "Near-injection Gridlist", and assigning the original grid index to the grid index of the near-injection region in step 5, and recording this reverse mapping result with the global grid list.

[0043] Step 113, create a container (or a large array, "Near-injection GridNum[Nnear]") to store the grid around each near-injection region. ( The number of adjacent grids within the range of the standard deviation of the Gaussian function is shown in Algorithm 2 in Table 2.

[0044] Table 2: Pseudocode for the algorithm to calculate the number of neighboring grids for each grid within the near-orifice region

[0045]

[0046] Here, "Tid" refers to the grid index of the mesh within the near-orifice area. It is the total number of grid cells in the near-orifice area. To determine using Algorithm 1 ( The grid position vector in the near-orifice region grid list. This is the current grid's position vector; Near-injection Gridlist [Tid] is the original index of the current grid obtained using the "Near-injection Gridlist" array of near-injection region grids from Algorithm 1. In step 3 of Algorithm 2, using... arrive Does the distance judgment satisfy the condition? If the conditions are met, then execute the atomic operation "atomicAdd" for neighbor grid counting in step 4. , The initial injection diameter is given by , and '&Near-injection GridNum[Tid]' represents the value of each near-field grid in the GPU parallel architecture. The number of neighboring grids within the range. "&" is the operator for taking the memory address of the array variable so as to provide the target memory location for the atomic operation of counting neighboring grids. "Near-injectionGridNum[]" is an integer array, and the "1" after it indicates that the counting step is 1.

[0047] Step 114: Calculate the total number of neighboring meshes in all near-orifice regions, denoted as . Create a container (or a large array named "Offset[]") to record the offset of the number of neighbors for each near-hole region mesh. Then, store each near-hole region mesh... The indices of all adjacent grids within the range are recorded in a new vector (or a contiguous array named "NearTid[]"), and the pseudocode is shown in Algorithm 3 in Listing 3 below. Here, the index "Tid" refers to the grid index of the near-orifice region.

[0048] Table 3: Pseudocode for the algorithm to calculate the total number of neighboring meshes in the near-orifice region

[0049]

[0050] Figure 2 The pseudocode implementation process is shown in the diagram. Algorithm 3 mainly includes: initializing the counter variable in step 1; and steps 2-5, traversing each grid within the near-orifice region on the CPU. Execute the storage grid in a loop The starting position of the neighbor list is determined, and the total number of neighbors is accumulated (to determine storage space), and an offset index table is constructed; Steps 6-14, the neighbor index is stored through GPU parallel processing, including: Steps 6-7, obtaining the storage starting position of the current mesh; Steps 8-14, determining the storage starting position based on the distance between the current mesh and the meshes in the vicinity of the nozzle. The process involves iterative neighbor determination and compact storage to ultimately obtain the total number of neighbor meshes for all near-orifice regions.

[0051] The characteristic of Algorithm 3 is that it uses CPU-GPU collaborative calculation of the offset to solve the problem of inconsistent neighbor numbers in each near-hole region grid caused by grid inhomogeneity. In this process, each thread writes to a contiguous memory block independently, without introducing atomic operation contention. The advantage is that in the large array NearTid[], the grid neighbors are grouped according to each near-field grid, that is, each group contains neighbors of a certain near-hole region grid, which can save memory and facilitate calling.

[0052] Step 110 establishes an improved pre-stored neighbor list strategy, which completes the most time-consuming neighbor grid search in the finite volume method under the GPU architecture at the preprocessing node, and stores the neighbor grid results in a large array in advance for later use, avoiding repeated neighbor grid searches in real-time computing and significantly reducing the retrieval burden in the real-time computing stage.

[0053] Step 120: Using the droplet size of the liquid and the pre-stored near-nozzle region grid neighbors in the large array, solve the gas phase NS equation to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature.

[0054] Step 130: Determine the droplet diameter using the Navier-Stokes equations for the frozen gas phase. The relationship between the droplet and the grid size (Δ) is determined using the following steps to obtain information about the grid surrounding the droplet (or the physical quantities of the flow field acting on the droplet by the gas phase). Then, the droplet governing equations are solved to obtain droplet information including droplet velocity, mass, and temperature. Specifically, this includes:

[0055] Step 131: When the droplet diameter At that time, the upstream averaging method is used to obtain information about the grid around the droplet, such as... Figure 3 As shown in (a), the upstream position of the droplet is determined specifically by the vector relationship between the droplet's motion direction and the airflow direction, including using the droplet diameter. By characterizing the droplet size, the distance from the upstream of the droplet is obtained. Grid within the area, The distance between the upstream region and the droplet center is given. Information about the surrounding mesh neighbors is obtained by directly calling the large array of pre-stored mesh neighbors in the near-orifice region from step 113. The basic flow field physics of the mesh surrounding the droplet are then calculated. The average value is taken to obtain the information of the grid around the droplet. The information of the grid around the droplet, including the flow field, is labeled using the index `seen`. The specific information of the grid around the droplet is given by the following formula:

[0056] (2)

[0057] in, ; It is upstream The number of grid cells within the area; , It is the gas phase density of the grid surrounding the droplet. These are the gas phase velocities in the three coordinate directions of the grid surrounding the droplet. It is the gas phase pressure of the grid surrounding the droplet. It is the gas phase temperature of the grid surrounding the droplet; information about the grid surrounding the droplet. Depend on Composition, in which, It is the gas phase density acting on the droplet. It is the gas phase velocity acting on the droplet. It is the gas phase pressure acting on the droplet. It is the temperature of the gas phase acting on the droplet.

[0058] Step 132: When In this case, the grid containing the droplet and the six surrounding grids are used, such as... Figure 3 As shown in (b), the inverse distance weighting method is used to obtain information about the mesh around the droplet, specifically given by the following formula:

[0059] (3)

[0060] in, It is the position vector of the grid surrounding the droplet. It is the position vector of the droplet; It is the normalized inverse distance coefficient between the grid surrounding the droplet and the droplet.

[0061] Step 133: Substitute the information of the surrounding mesh of the droplet, such as gas phase velocity, gas phase density, gas phase pressure, and gas phase temperature obtained in steps 131 and 132, into the following liquid phase control equations: the droplet position equation given by equation (4), the droplet force equation given by equation (5), and the breakup equation (see: Li Peibo. Numerical simulation of mixing and combustion process of transverse spray in supersonic airflow [D]. Changsha: National University of Defense Technology, 2019: 43-48), and use the Runge-Kutta method to solve for the droplet velocity. ,quality and temperature The governing equations for the liquid phase include:

[0062] (4)

[0063] (5)

[0064] in, For time, It is the density of the droplet; It is the drag coefficient. It is the gas phase velocity acting on the droplet.

[0065] Step 140, calculate the source terms carried on each grid, specifically including:

[0066] Step 141: Apply the droplet velocity obtained in Step 130 ,quality ,temperature Substituting into formula (1), the quality source term is calculated using the source term regression kernel function. Momentum source terms and energy source items .

[0067] (6)

[0068] (7)

[0069] (8)

[0070] in, It is the relative position vector between the droplet and the center of the grid. Let V be the grid volume where the droplet is located. For particle pack coefficients, The latent heat of the droplet; This is the source term regression kernel function.

[0071] Step 142: Considering the spray as a multi-droplet system, here we propose a temporary nearest-node kernel function:

[0072] (9)

[0073] That is, the source terms of all droplets in a grid are added to their respective grids using the nearest node kernel function.

[0074] Step 143: Calculate the source term load for the current mesh. ; Determine if the source term load Greater than a given threshold (For example The source terms are assigned to the surrounding grid using the Gaussian kernel function given by formula (10).

[0075] (10)

[0076] ;

[0077] in, Gaussian function The standard deviation.

[0078] The aforementioned nearest node kernel function With Gaussian kernel function The differences in the given source item allocation methods and effects are as follows: Figure 4 As shown. Among them, Figure 4(a) shows the source term allocation form and effect given by the nearest node kernel function. The influence of the droplet center is completely and only allocated to the grid node closest to its geometric position (indicated by the thick arrow pointing to the node in the figure). In terms of allocation effect, the influence of the source term is highly concentrated on the grid node closest to the droplet, thus generating strong local perturbations near the node. Figure 4 (b) shows the source term allocation form and effect given by the Gaussian kernel function. The figure illustrates that the influence of the droplet center is allocated to multiple surrounding grid nodes according to the law of a Gaussian function. In terms of allocation effect, the influence of the source term is smoothly distributed to multiple grid nodes within a certain range around the droplet. Therefore, the disturbance generated by the source term is diffused into a finite region, avoiding excessive concentration on a single node. This smooth allocation method significantly suppresses non-physical pressure oscillations. The information of the surrounding grid neighbors used in the aforementioned process can be directly obtained by calling the large array of pre-stored grid neighbors in the near-orifice region from step 110.

[0079] Step 150: After obtaining the source terms on each grid in step 140, the frozen gas-phase Navier-Stokes equations are decomposed using the Eulerian method. The flow field of the frozen gas-phase Navier-Stokes equations is then corrected using source terms, and the basic flow field physical quantities for the next time step are updated. .

[0080] Specifically, the flow field expression after freezing the gas phase NS equations is as follows:

[0081] (11)

[0082] Then, by decomposition using the Euler method, we obtain:

[0083] (12)

[0084] in, It is the current moment. It's the next moment;

[0085] Solving the above equation yields the conserved flow field quantities at the next time step:

[0086] (13)

[0087] Update the basic flow field physical quantities for the next time step. .

[0088] Step 160: Repeat steps 120 to 150 to continuously update the basic flow field physical quantities, droplet physical quantities, and the new basic flow field physical quantities after the droplet corrects the flow field, until the updated basic flow field physical quantities at all times are obtained.

[0089] The present invention provides a numerical simulation method for supersonic liquid jet atomization. The overall approach is to solve the supersonic two-phase flow process using a decoupling algorithm. This method is based on the Euler-Lagrange method for bidirectional coupling calculation of the gas-liquid two-phase flow, where the gas-phase flow control equations and droplet source terms are solved separately to achieve numerical simulation of supersonic liquid jet atomization. The basic idea of ​​the decoupling algorithm is to first solve the gas-phase flow field without droplet influence, then freeze the gas-phase flow and solve the droplet phase control equations, and statistically analyze the source term effect of droplets on the gas phase. The specific steps of the bidirectional coupling calculation based on the Euler-Lagrange method are as follows: at each time step, the gas-phase flow field is first calculated; then, based on the updated gas-phase flow field information, the motion and breakup equations of the droplets are solved, thereby updating the liquid mist field information. Simultaneously, at the end of the liquid phase time step, the source term effect generated by the droplets when droplets are present in the gas-phase control body is considered. When the time step of the liquid phase calculation reaches the same time step as the gas phase calculation, a single gas-liquid cycle calculation is completed. The above steps are then repeated at the next time step. The specific calculation process is as follows: Figure 5 As shown.

[0090] This invention proposes an improved numerical simulation method for supersonic liquid jet atomization in the above embodiments, aiming to solve the numerical instability and jet-mesh scale dependence problems of the traditional Eulerian-Lagrange method in scenarios simulating droplets larger than the mesh. First, an improved pre-stored neighbor list strategy is proposed, specifically addressing the large amount of neighbor mesh search work required by the finite volume method under GPU architecture. Centralized computation in the preprocessing stage significantly reduces the retrieval burden in the real-time computation stage. Second, considering the characteristics of multi-droplet systems in spray fields, an innovative source term allocation scheme combining the nearest node kernel function and the Gaussian kernel function is adopted. Source terms are first allocated to the nearest mesh node and then diffused to the surrounding mesh within a certain range, avoiding the instability of large droplet jet simulation caused by excessive concentration of source terms. Finally, an upstream averaging method is introduced to solve the problem of underestimated penetration depth caused by droplet self-perturbation in classical methods. This invention significantly saves computational resources by preprocessing a large amount of neighbor mesh search work, achieving stable simulation of jets of arbitrary size with reasonable simulation error. Meanwhile, it eliminates the dependence of numerical injection size on the grid and reduces empirical errors caused by manually setting injection parameters, making it suitable for a wide range of engineering applications, especially for numerical simulation of complex flow fields such as supersonic combustion chambers.

[0091] The present invention provides an experimental verification of the numerical simulation method for supersonic liquid jet atomization provided in the above embodiments. The specific configuration and inflow conditions are based on a supersonic transverse jet experiment conducted in a supersonic wind tunnel with a designed Mach number of 2.1. The stagnation temperature and stagnation pressure were 300 K and 891 kPa, respectively. In the specific experiment, water was used as the test liquid, with a density of 998 kg / m³. The gas-liquid momentum ratio was defined as 3.9, and the injection flow rate was approximately 29.8 g / s. Simulation computational domain: size [not specified]. Around the nozzle The area within this region is called the jet near-orifice region. , and The mesh was refined in the direction, where and Grid length in direction and They are 0.12mm respectively. Grid length in direction ( At the wall surface, it is equal to 0.01 mm. At room temperature, at time intervals... Discrete droplets are injected, as shown in Figure 6. The nozzle is located at (50, 0, 30) on the wall. Three injection diameters are used. The actual initial nozzle diameter is The grid size where the droplets are located at the injection point:

[0092] .

[0093] Different bidirectional coupling methods were used, with three sets of calculation examples designed for each method. For example, method a uses the classic point source assumption bidirectional coupling method, designing three sets of calculation examples (a1, a2, a3). Method b uses the improved method provided in this invention, designing three sets of calculation examples (b1, b2, b3). The specific calculation example settings are shown in Table 4. The results show that the classic point source assumption bidirectional coupling method cannot calculate large jet diameters (e.g., , The penetration depth error when taking values ​​of a2 and a3 (corresponding to a values ​​of 0.4 and 1), and the small injection diameter (e.g.) , In example a1 with a depth of 0.1, although convergence was achieved, the traditional depth error was relatively high (7.02%). After using the improved bidirectional coupling method, regardless of... The calculations showed stable and convergent results, with the penetration depth error reduced to below 4%. A comparison of the quantitative penetration depth results with experimental data is provided. Figure 7 As shown, where, The vertical axis represents the penetration depth. The x-axis is The method provided in this invention was used to obtain three sets of droplet distribution calculation results, such as... Figure 8 As shown.

[0094] Table 4: Simulation Case Settings

[0095]

[0096] In one embodiment, the present invention provides a numerical simulation device for supersonic liquid jet atomization, which implements the steps of the aforementioned numerical simulation method for supersonic liquid jet atomization. The device includes:

[0097] The first module is used to preprocess the gas-phase Navier-Stokes equations for supersonic liquid jet atomization, including using an improved pre-stored neighbor list strategy to establish a large array to pre-store the grid neighbors of the near-orifice region.

[0098] The second module is used to solve the gas phase Navier-Stokes equations using the liquid droplet size and the pre-stored near-nozzle region grid neighbors in a large array, to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature;

[0099] The third module is used to freeze the Navier-Stokes equations in the gas phase. It uses the relationship between the droplet diameter and the grid size where the droplet is located to obtain information about the grid around the droplet. The information about the grid around the droplet is substituted into the liquid phase control equations, and the Runge-Kutta method is used to solve the liquid phase control equations.

[0100] The fourth module is used to calculate the source terms carried on each grid.

[0101] The fifth module is used to correct the flow field given by the frozen gas-phase Navier-Stokes equations using source terms on each grid, and update the basic flow field physical quantities for the next time step.

[0102] The sixth module is used to repeatedly execute the steps of the second to fifth modules until the updated basic flow field physical quantities at all times are obtained.

[0103] On the other hand, the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the numerical simulation method for supersonic liquid jet atomization provided in any of the above embodiments. The computer device can be a server. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store sample data. The network interface of the computer device is used for communication with external terminals via a network connection.

[0104] In addition, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the numerical simulation method for supersonic liquid jet atomization provided in any of the above embodiments.

[0105] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0106] Matters not covered in this invention are common knowledge.

[0107] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0108] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0109] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A numerical simulation method for supersonic liquid jet atomization, characterized in that, include: Step 110 involves preprocessing the gas-phase Navier-Stokes equations for supersonic liquid jet atomization, including using an improved pre-stored neighbor list strategy to establish a large array of pre-stored grid neighbors for the near-orifice region; including: Step 111: Initialize the variables in the gas phase Navier-Stokes equations, perform mesh coordinate transformation, and perform mesh preprocessing for computation arrays and memory allocation; Step 112: Using an improved pre-stored neighbor list strategy, filter out areas falling within the near-orifice region. The grid index is recorded in the near-orifice area grid list; Step 113, create a large array to store the number of adjacent grids around each near-orifice region grid; including: Create an array of near-injection gridlist[] and obtain the original index of the current grid; Calculate the grid position vector in the near-orifice region grid list. Position vector to the current grid distance And determine whether the distance judgment condition is met. ,in, , This is the initial injection diameter; If satisfied Then, atomic operations are performed to obtain the number of neighboring grids for each grid. This number is stored in a large array `Near-injection GridNum[]` under a GPU parallel architecture, representing the number of neighboring grids around each near-injection region. The number of adjacent grids within the range; If not satisfied Then, repeat step 113 for the next grid. Step 114: Calculate the offset of the number of neighbors of each near-hole region mesh using a CPU-GPU collaborative method, and calculate the total number of neighbor meshes for all near-hole regions. Step 120: Using the droplet size of the liquid and the pre-stored near-nozzle region grid neighbors in the large array, solve the gas phase Navier-Stokes equations to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature; Step 130: Freeze the gas phase Navier-Stokes equations, and obtain information about the grid around the droplet by using the relationship between the droplet diameter and the grid size where the droplet is located; substitute the information about the grid around the droplet into the liquid phase control equations, and solve the liquid phase control equations using the Runge-Kutta method; Step 140: Calculate the source terms carried on each grid. Step 150: Use the source terms on each grid to correct the flow field given by the frozen gas-phase Navier-Stokes equations and update the basic flow field physical quantities for the next time step. Step 160: Repeat steps 120 to 150 until the updated basic flow field physical quantities at all times are obtained.

2. The numerical simulation method for supersonic liquid jet atomization according to claim 1, characterized in that, The gas-phase Navier-Stokes equations for supersonic liquid jet atomization can be expressed in the following form with droplet source terms: ； in, For the incoming Reynolds number, For flow field conservation quantities, For gas phase density, These are the gas phase velocities in three coordinate directions. It is gas phase energy; , , Coordinate systems Middle Inviscid flux in the direction, , , for Viscous flux in the direction; Indicates the source term, It is a quality source item. It is a momentum source term. It is an energy source term. This indicates the transpose of a vector or matrix; use Represents the basic flow field physical quantities, where, It is gas phase pressure. It is the gas phase temperature.

3. The numerical simulation method for supersonic liquid jet atomization according to claim 2, characterized in that, Step 112 includes: The operation location for storing indexes in integer variables is defined using atomic operations based on GPU parallel computing; Define a large array Near-injection Gridlist[] to record and store the regions that fall near the injection port. Grid index.

4. The numerical simulation method for supersonic liquid jet atomization according to claim 3, characterized in that, Step 114 includes: Initialize the counter variable; On the CPU, each grid within the near-orifice region is traversed, the starting position of the neighbor list of the stored grid is executed in a loop, the total number of neighbors is accumulated, and an offset index table is constructed. Neighbor index storage is performed using GPU parallel processing, including: obtaining the starting storage position of the current grid; and determining the storage condition based on the distance between the current grid and other grids in the vicinity of the nozzle. The process involves iterative neighbor determination and compact storage to ultimately obtain the total number of neighbor meshes for all near-orifice regions.

5. The numerical simulation method for supersonic liquid jet atomization according to claim 4, characterized in that, Step 130 includes: Step 131, let It is the grid size where the droplet is located, when the droplet diameter At that time, the upstream position of the droplet is determined by the vector relationship between the direction of droplet motion and the direction of airflow, including: Using droplet diameter By characterizing the droplet size, the distance from the upstream of the droplet is obtained. Grid within the area, The distance between the upstream region and the center of the droplet; Using information from the mesh neighbors in the near-orifice region, the basic flow field physical quantities of the mesh surrounding the droplet are calculated. After averaging, the information of the grid around the droplet is obtained: ； in, ; It is upstream The number of grid cells within the area; These represent the basic flow field physical quantities of the grid surrounding the droplet. It is the gas phase density of the grid surrounding the droplet. These are the gas phase velocities in the three coordinate directions of the grid surrounding the droplet. It is the gas phase pressure of the grid surrounding the droplet. It is the gas phase temperature of the grid surrounding the droplet; information about the grid surrounding the droplet. Depend on[ Composition, in which, It is the gas phase density acting on the droplet. It is the gas phase velocity acting on the droplet. It is the gas phase pressure acting on the droplet. It is the temperature of the gas phase acting on the droplet; Step 132, when the droplet diameter In this case, the information about the grid surrounding the droplet is obtained by taking the inverse distance-weighted average of the grid where the droplet is located and the six surrounding grids: ； in, It is the position vector of the grid surrounding the droplet. It is the position vector of the droplet; It is the normalized inverse distance coefficient between the grid surrounding the droplet and the droplet; Step 133: Substitute the information of the grid around the droplet into the liquid phase control equations and solve them using the Runge-Kutta method to obtain the droplet velocity. droplet mass and droplet temperature The liquid phase control equations include the droplet position equation, the droplet force equation, and the breakup equation.

6. The numerical simulation method for supersonic liquid jet atomization according to claim 5, characterized in that, Step 140 includes: Step 141, droplet velocity droplet mass and droplet temperature Substituting into the gas-phase Navier-Stokes equations, and using the source term regression kernel function... Calculate quality source items Momentum source terms and energy source items : ； ； ； in, It is the relative position vector between the droplet and the center of the grid. Let V be the grid volume where the droplet is located. For particle pack coefficients, For time, The latent heat of the droplet; Step 142, define the temporary nearest node kernel function: ； Add the source terms of all droplets in the current grid to their respective grids using the nearest node kernel function; Step 143, using the threshold condition of the source term load in the current mesh and the Gaussian kernel function, distribute the source term to the surrounding mesh, including: Calculate the source term load: ， in, It is the density of the droplet; Set the threshold for source load. ; like The source term is then distributed onto the grid surrounding the droplet using the following Gaussian kernel function: ； in, ; Gaussian function Standard deviation; like If so, then jump to the next moment.

7. The numerical simulation method for supersonic liquid jet atomization according to claim 6, characterized in that, Step 150 includes: The flow field expression obtained after freezing the gas phase Navier-Stokes equations is as follows: ； The above flow field expression is decomposed using the Euler method to obtain the decomposed flow field expression: ； in, Indicates the current moment. Indicates the next moment; Solving the decomposed flow field expression, we obtain the conserved flow field quantities at the next time step: ； Update the basic flow field physical quantities for the next time step. .

8. A numerical simulation device for supersonic liquid jet atomization, characterized in that, The apparatus is used to implement the steps of the numerical simulation method for supersonic liquid jet atomization as described in any one of claims 1-7, wherein the apparatus comprises: The first module is used to preprocess the gas-phase Navier-Stokes equations for supersonic liquid jet atomization, including using an improved pre-stored neighbor list strategy to establish a large array to pre-store the grid neighbors of the near-orifice region. The second module is used to solve the gas phase Navier-Stokes equations using the liquid droplet size and the pre-stored near-nozzle region grid neighbors in a large array, to obtain the basic flow field physical quantities at the current moment; the basic flow field physical quantities include at least gas phase density, gas phase velocity, gas phase pressure and gas phase temperature; The third module is used to freeze the Navier-Stokes equations in the gas phase. It uses the relationship between the droplet diameter and the grid size where the droplet is located to obtain information about the grid around the droplet. The information about the grid around the droplet is substituted into the liquid phase control equations, and the Runge-Kutta method is used to solve the liquid phase control equations. The fourth module is used to calculate the source terms carried on each grid. The fifth module is used to correct the flow field given by the frozen gas-phase Navier-Stokes equations using source terms on each grid, and update the basic flow field physical quantities for the next time step. The sixth module is used to repeatedly execute the steps of the second to fifth modules until the updated basic flow field physical quantities at all times are obtained.

Images