A method for generating high-precision simulation inflow conditions for reactor thermal fluids

Abstract

The present invention discloses a method for generating high-precision simulation inflow conditions for reactor thermal fluids, belonging to the technical field of computational fluid dynamics. The present invention utilizes experimental data in the turbulence database and fully developed turbulence structure data obtained through two high-precision simulation methods, DNS and LES. When numerically simulating reactor fluids, according to the calculation requirements and model design, turbulence data under different working conditions and different physical conditions are selected, and this data is matched to the flow field grid being calculated using the interpolation method to generate inflow conditions with turbulent structures. The goal of the proposed method is to generate inflow conditions that can be used for high-precision simulation of reactor thermal fluids, have physical properties and fully developed turbulent structures, meet the simulation calculation requirements of users, do not require users to make additional adjustment designs at the inlet, and save time and computational costs.

Description

Technical Field

[0001] The present invention relates to the technical field of computational fluid dynamics, and particularly relates to a method for generating high-precision simulation inflow conditions for reactor thermal fluids. Background Art

[0002] How to accurately describe and simulate the complex flow characteristics inside a nuclear reactor is a key issue in the field of nuclear energy engineering. As an important facility for energy generation, the internal fluid dynamics process of a nuclear reactor directly affects the cooling effect of fuel elements, the performance of heat exchangers, and safety, etc. The inflow conditions of the fluid play a crucial role, which directly affect the formation of turbulent structures, the efficiency of energy transfer, and the stability of the system. Computational Fluid Dynamics (CFD) is an important method for simulating physical processes such as fluid flow and heat transfer in a reactor. By solving the Navier-Stokes equations to simulate and predict the fluid flow process, and the inflow conditions define the initial state of the fluid flow problem, which is one of the necessary conditions to ensure a unique solution of the Navier-Stokes equations. Accurate inflow conditions can ensure the stability and convergence of the numerical solution, and avoid unstable or divergent phenomena caused by initial errors. Therefore, it is crucial to deeply study the inflow conditions of reactor thermal fluids.

[0003] Different thermal fluid simulation methods have different sensitivities to inflow conditions. The commonly used Reynolds-averaged Navier-Stokes (RANS) method in engineering models all turbulent fluctuations, ignoring the influence brought by the fluctuations, and has low requirements for boundary conditions; the Large-eddy Simulation (LES) method only simulates large-scale eddies and models smaller-scale eddies, and is more sensitive to inflow conditions; Direct Numerical Simulation (DNS) directly simulates all scales of turbulence, and small perturbations can continue to propagate over time and space, and is even more sensitive to inflow conditions. To improve the accuracy of the simulation method for reactor thermal fluids, high-quality and physically realistic inflow conditions are essential because they affect the flow of downstream fluids and thus the accuracy of the entire simulation. If the fluctuations of the inflow turbulence are not real and reasonable enough, the turbulence fluctuations may dissipate before reaching the computational region, or it may take a long distance to form realistic turbulence.

[0004] At present, the commonly used methods for setting the inflow conditions of reactor thermal fluids are as follows: specifying a steady flow rate or velocity field, using data obtained from laboratory measurements as the inflow conditions, and using the internal flow field distribution obtained from pre - conducted numerical simulations as the inflow conditions, etc. However, due to the geometric complexity, fine dimensions, large scale, and high - fidelity numerical calculation requirements of the reactor core structure, the existing methods have the following deficiencies: 1) The method of specifying a steady flow rate or velocity field completely ignores turbulent fluctuations, cannot accurately simulate the dynamic changes of the fluid over time under different working conditions or operating stages, and cannot capture the local non - uniformity caused by factors such as geometric shape, component layout, and local heat transfer. This method is only applicable to computational simulations that require simplified treatment of boundary condition changes; 2) Directly using the data obtained from laboratory measurements as the inflow conditions has the problem that the fluid conditions in the laboratory may not exactly match those in the nuclear reactor, such as differences in the size of the model; 3) For the method of pre - conducting numerical simulations, when facing reactors of large scale, fine grids, and high - precision computational simulations, different geometric models and boundary conditions require pre - simulation calculations, which will significantly increase the simulation time and computational cost.

[0005] To improve the calculation accuracy, some scholars have proposed to use the existing data and calculation results to generate the inflow conditions with turbulent intensity and fluctuations. For example, Dhamankar et al. (Dhamankar, N.S., G.A. Blaisdell, and A.S. Lyrintzis. Overview of Turbulent Inflow Boundary Conditions for Large-Eddy Simulations[J]. AIAA Journal 56(4): 1317–1334. doi:10.2514 / 1.J055528.) and Tabor et al. (G.R. Tabor, M.H. Baba-Ahmadi. Inlet conditions for large eddy simulation: A review[J]. Computers & Fluids. 2010: 553-567.) summarized the conclusions that the turbulent database can provide detailed and accurate turbulent inflow conditions, has wide applicability, and the calculation results have high accuracy and reliability. Combining with the work of J.U. Schluter et al. (Schluter, Jorg & Moin, Parviz & Pitsch. Large-Eddy Simulation Inflow Conditions for Coupling with Reynolds-Averaged Flow Solvers[J]. Aiaa Journal-AIAA. 2004 42(3) 478-484. 10.2514 / 1.3488.) on the application analysis of the turbulent database in LES inflow conditions, it is found that its calculation results are consistent with the experimental data.

[0006] Based on the deficiencies of the existing technology and combined with the existing research on the turbulent database, the present invention proposes a method for generating inflow conditions for high-precision simulation of reactor thermal fluids. Summary of the Invention

[0007] The purpose of the present invention is to propose a method for generating inflow conditions for high-precision simulation of reactor thermal fluids to solve the problems raised in the background technology. The present invention pre-stores the experimental data and the results of other simulations calculated by RANS, LES, and DNS methods, including flow field information, grid point coordinates, velocity components in each direction, and the solved passive scalar information, into the database. When performing numerical simulations, according to the calculation requirements, select and read the corresponding turbulent data files, and use the interpolation method to match them to the flow field grid positions to be solved to generate turbulent inflow conditions.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for generating high-precision simulation inflow conditions for reactor thermal fluids, comprising the following steps:

[0010] S1. Construct a turbulence database, which includes a data source module, a data storage module, a data reading module, and a data application module;

[0011] S2. Obtain hydrodynamic experimental data, and use the RANS, LES, and DNS methods to obtain simulation calculation results, and store the obtained experimental data and simulation results in the data source module;

[0012] S3. Divide the experimental data set simulation results obtained in S2 into a flow field information file and a turbulence data file, and store them in the data storage module. When storing, unify the data type and length to improve the storage efficiency and reading speed;

[0013] S4. Based on the data reading module, use multi-threading and parallel I / O technology to read data to improve the reading efficiency;

[0014] S5. After the data reading is completed, before applying the data to the flow field, prepare the network of the flow field to be simulated and set the boundary conditions. Based on the data application module, determine whether the turbulence data matches the grid or spatial resolution of the flow field. If it matches, directly enter S6; if it does not match, based on the geometric classification-based adaptive interpolation algorithm, according to the geometric dimension of the model and the structured and unstructured nature of the grid, apply the turbulence data to different flow field models, and then enter S6;

[0015] S6. Apply the input data for simulation calculation;

[0016] S7. After the simulation calculation is completed, compare and verify the calculation results with the experimental data or known theoretical results, evaluate the accuracy and applicability of the simulation results, and adjust the simulation parameters or the application method of the data according to the verification results.

[0017] Preferably, the flow field information file in S3 is used to store flow field and turbulence information, including turbulence energy spectrum, correlation time scale, turbulence intensity, type of calculation process, and flow field information; the turbulence data file is used to store data sets, including point coordinates, velocity component values in each direction, and passive scalar values.

[0018] Preferably, the use of multi-threading and parallel I / O technology to read data in S4 specifically includes the following content:

[0019] S4.1. Data chunking: Divide the large file into multiple small chunks, and each chunk contains a certain number of flow field data points;

[0020] S4.2, Multi-threaded Reading: Use the standard thread library to create multiple threads, and each thread is responsible for reading one or more data blocks;

[0021] S4.3, Data Buffering: Each thread stores the read data blocks into a shared buffer, and uses a lock mechanism to ensure thread safety;

[0022] S4.4, Data Parsing: The read data blocks are parsed into specific flow field point data in the buffer, including coordinates, velocity components, and passive scalars.

[0023] Preferably, for the adaptive interpolation algorithm based on geometric classification described in S5, according to the geometric dimension of the model, the structured and unstructured nature of the grid, the turbulent data is applied to different flow field models, and the specific content is as follows:

[0024] If the model is one-dimensional, it means that the flow field modeling is simple, and the linear interpolation method (Linear Interpolation) is used to apply the data;

[0025] If the model is two-dimensional, the networks used in the flow field model are divided into two types: structured and unstructured. When the grid constructed for the flow field is a structured grid, the bilinear interpolation method (Bilinear interpolation) is used for physical quantity interpolation; if the constructed grid is an unstructured grid, the high-precision interpolation method based on WENO (Weighted Essentially Non-Oscillatory-based high-accuracy interpolation) is used for interpolation;

[0026] If the model is three-dimensional, the radial basis function interpolation method (Radial basis function interpolation) is used to insert the turbulent data.

[0027] Compared with the prior art, the present invention provides a method for generating inflow conditions for high-precision simulation of reactor thermal-hydraulics, having the following beneficial effects:

[0028] (1) Saving time cost and computing resources: Facing the numerical simulation requirements of large-scale reactors, high-fine grids, and small time steps, the present invention, by using a pre-generated turbulent database, does not need to repeat large-scale simulation calculations, directly extracts the inflow conditions from the database, and significantly reduces the computing time and resource consumption.

[0029] (2) Providing real turbulent structures: The turbulent database is based on a large amount of experimental data and high-precision numerical simulation results. The present invention can accurately describe the turbulent characteristics, such as turbulent energy spectrum, correlation time scale, turbulent intensity, etc., to ensure the authenticity and accuracy of the inflow conditions.

[0030] (3) It can be applied to the high-precision simulation of reactor thermal-hydraulic fluids: The data in the turbulence database is sourced from experimental data or high-precision numerical simulations. The present invention can provide relatively real and accurate turbulence characteristics, with a certain degree of verification and reliability, and can generate high-quality inlet conditions for the high-precision simulation of reactor thermal-hydraulic fluids.

[0031] (4) Wide applicability: The turbulence database constructed in the present invention can cover various fluid types and different operating conditions, such as fluid types with different flow velocities, temperatures, pressures, etc., including steady-state conditions, transient conditions, and accident conditions. The database can adapt to various nuclear reactor design and operating conditions, with strong generality and adaptability. Description of the Drawings

[0032] Figure 1 This is the overall schematic diagram of a method for generating inlet conditions for high-precision simulation of reactor thermal-hydraulic fluids proposed in Embodiment 1 of the present invention, mainly including: saving the experimental data or the results obtained by simulating and calculating using the RANS, LES, and DNS methods into the turbulence database. When conducting the main simulation, select appropriate turbulence data from the database and apply it to the grid points using the interpolation method, that is, generate the inlet conditions with turbulent structures.

[0033] Figure 2 This is the schematic diagram of the four component modules of the constructed turbulence database mentioned in Embodiment 1 of the present invention, including: data source module, data storage module, data reading module, and data application module;

[0034] Figure 3 This is the specific method flowchart of a method for generating inlet conditions for high-precision simulation of reactor thermal-hydraulic fluids proposed in Embodiment 1 of the present invention. Detailed Implementation Manner

[0035] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0036] Embodiment 1:

[0037] To better adapt to the three-dimensional geometric structure of complex helices, a hybrid mesh generation method is often adopted, which combines various mesh forms such as tetrahedrons, hexahedrons, and triangular prisms. In addition, when studying the tangential plane flow characteristics of the geometric structure of complex helices, a two-dimensional hybrid mesh generation method is also often used, which combines triangular and quadrilateral mesh forms and is divided according to the specific characteristics of each part in order to achieve the best simulation effect. To enable the hybrid meshes of tetrahedrons (hereinafter referred to as tetra), hexahedrons (hereinafter referred to as hex), triangular prisms (hereinafter referred to as wedge) in three dimensions and triangles (hereinafter referred to as tri) and quadrilaterals (hereinafter referred to as quad) in two dimensions to perform spectral element method DNS and solve the problems of non-automation, long time consumption, and high memory requirements in the conversion process, the present invention proposes a method for generating inflow conditions for high-precision simulation of reactor thermal fluids. Through parallel processing and optimization algorithms, the hybrid meshes of complex flow fields of helical structures are effectively converted into high-quality hexahedral meshes, thus overcoming the automation problems and high resource consumption problems in traditional methods. This technology can not only handle the complex hybrid meshes commonly found in complex flow fields of helical structures, but also significantly improve the conversion efficiency and reduce the demand for computing resources while maintaining grid consistency, providing an efficient and reliable grid generation solution for DNS based on the spectral element method.

[0038] The following will explain a method for generating inflow conditions for high-precision simulation of reactor thermal fluids proposed by the present invention in conjunction with the accompanying drawings and specific examples, which specifically includes the following contents.

[0039] Example 1:

[0040] Please refer to Figure 1-2 The present invention proposes a method for generating inflow conditions for high-precision simulation of reactor thermal fluids, which consists of 4 modules as shown in Figure 2 the following figure.

[0041] Module 1. Data source: To ensure that the generated inflow conditions are physically reasonable, the data in the turbulence database comes from experimental data and the calculation results of other simulations. The experimental data mainly comes from fluid mechanics experiments under strictly controlled laboratory environments. These experiments use advanced technologies such as PIV (Particle Image Velocimetry) and LDV (Laser Doppler Velocimetry) to obtain high-precision velocity field and turbulence characteristic data. Other high-precision simulation calculation results refer to using RANS, LES, and DNS methods, using high-quality and fine meshes, and simulating the reactor thermal fluids to obtain high-fidelity results. To improve the universality of the turbulence database in the field of high-precision simulation of reactor thermal fluids, the database stores turbulence data under different working conditions, including steady-state conditions, transient conditions, and accident conditions, and provides different fluid types, such as different flow velocities, temperatures, pressures, etc.

[0042] Module 2. Data Storage: When storing turbulent data, strict naming conventions and data formats are adopted. The file name of each data file contains key flow field information, such as operating condition type, flow velocity, temperature, and pressure, etc., for easy retrieval and use. At the same time, a.md file is included in the directory where each data file is located, which details the flow field information of the data file, including turbulent energy spectrum, correlation time scale, turbulent intensity, type of calculated flow field, fluid information, etc. The data file stores the coordinates of each point in the flow field, the velocity components (u, v, w), and the values of passive scalars (such as temperature, concentration, etc.). These data are stored in binary format to improve storage efficiency and reading speed. Each data file is compressed and verified to ensure the integrity and accuracy of the data.

[0043] Module 3. Data Reading: The data in the turbulent database comes from laboratory data or high-precision LES and DNS calculations, and the amount of stored data is large. To efficiently read this data, multi-threading and parallel I / O technologies are adopted. The data reading process is divided into several steps: data chunking, multi-threaded reading, data buffering, and data parsing.

[0044] Module 4. Data Application: Before applying the data in the turbulent database to the flow field, the grid of the flow field to be simulated needs to be prepared and the boundary conditions need to be set. After selecting and reading the turbulent data according to the simulation model and calculation conditions, it is applied to the flow field grid. When the data does not match the simulation calculation grid spatially, through an adaptive interpolation algorithm based on geometric classification, according to the geometric dimensions of the model and the structured and unstructured nature of the grid, the turbulent data is applied to different flow field models, making the turbulent database highly versatile. The non-uniformity of the grid and the details of the turbulent structure are considered during the interpolation process to ensure the physical rationality and calculation accuracy of the interpolation results.

[0045] Based on the above content, the present invention further specifically describes the method for generating the inflow conditions for high-precision simulation of reactor thermal-hydraulic fluids proposed:

[0046] When storing turbulent data, relevant information of the flow field and turbulent numerical files need to be stored in each data folder. The file naming contains key information of the flow field, such as operating condition type, flow velocity, temperature, and pressure, etc. The file types and their descriptions are as follows:

[0047]

[0048] When in the data storage format, unify the data type and length for easy use in simulation calculations.

[0049] The specific method process is as Figure 3 shown, and the specific content is as follows:

[0050] (1) Understanding of data information

[0051] Before using the turbulence data, the user needs to understand the operating conditions, fluid condition information, and stored content of each data for later use. In addition, the user also needs to understand the data types and specific content included in the turbulence database, such as turbulence energy spectra, correlation time scales, turbulence intensities, etc. These data describe the spatial and temporal characteristics of turbulence, thus helping the user select the most suitable turbulence data according to their simulation calculation requirements.

[0052] (2) Preparation for simulation calculation

[0053] Before performing the flow field simulation, the user needs to prepare the mesh and initial condition settings required for the simulation, including defining the geometric shape of the computational domain, the initial flow field state, and boundary condition settings, and at the same time select the model used in the simulation, such as LES or DNS.

[0054] (3) Select appropriate turbulence data

[0055] The user selects turbulence characteristic data that meets the conditions and is physically reasonable according to the defined geometric model and calculation requirements.

[0056] (4) Read data

[0057] After completing the selection of turbulence data, read the file according to the file name of the data. Since there is a large amount of turbulence data information, in order to improve the reading efficiency and reduce time, a multi-threaded and parallel I / O method is used to read the data. Specifically, the data reading process is divided into the following steps:

[0058] Data chunking: Divide the large file into multiple small chunks, each chunk containing a certain number of flow field data points.

[0059] Multi-threaded reading: Use the standard thread library to create multiple threads, and each thread is responsible for reading one or more data chunks.

[0060] Data buffering: Each thread stores the read data chunk into a shared buffer, and uses a locking mechanism (such as a mutex) to ensure thread safety.

[0061] Data parsing: The read data chunk is parsed into specific flow field point data in the buffer, including coordinates, velocity components, and passive scalars.

[0062] The parallel I / O reading method is shown in the following pseudocode.

[0063]

[0064] Each processor first calculates the specific indices of the turbulent data files it is responsible for reading, and then each processor locates and reads the data blocks in the corresponding data files according to these indices. After reading the data, each processor parses it into flow field point data and preprocesses the parsed local data, including data compression and packaging, to optimize the subsequent data aggregation process. Subsequently, it is aggregated to the specified master processor (P0) using non-blocking parallel I / O operations, and P0 unpacks the data to form complete flow field data.

[0065] (5) Data grid interpolation matching

[0066] When applying the data to the simulated flow field, it is necessary to consider whether the turbulent data matches the grid or spatial resolution of the flow field.

[0067] If the selected data matches the grid and spatial resolution of the flow field, the data can be directly applied to the flow field for calculation.

[0068] If the data in the turbulent database does not match the grid or spatial resolution used in the actual simulation, interpolation method needs to be used to insert the data provided by the turbulent database into the corresponding positions in the simulation grid. For better generality and applicability of the turbulent database, the present invention proposes an adaptive interpolation algorithm based on geometric classification. The algorithm pseudocode is shown as follows.

[0069]

[0070] The following are the input parameters of the algorithm and their descriptions:

[0071]

[0072] In addition, the output content of the algorithm is the grid data with interpolation completed.

[0073] The design idea of this algorithm is: according to the input parameters, that is, the geometric dimensions of the model to be solved and whether the grid is structured, apply the turbulent data to the corresponding positions of the grid using interpolation method.

[0074] · If the model is one-dimensional, it means that the flow field modeling is relatively simple, and the data can be applied using linear interpolation.

[0075] Interpolation).

[0076] · If the model is two-dimensional, the grids used in the flow field model can be structured and unstructured. When structured grids are constructed for the flow field, due to the regularity of structured grids, the relatively simple bilinear interpolation method is used for physical quantity interpolation. If the grids are unstructured, a high-precision interpolation method based on WENO (Weighted Essentially Non-Oscillatory-based high-accuracy interpolation) is used for interpolation.

[0077] · When the model is three-dimensional, more refined unstructured grids are used in the flow field to construct the model to improve the calculation accuracy. Therefore, when inserting turbulent data, the radial basis function interpolation method (Radial basis function interpolation) is used to meet the requirements of high-precision simulation calculations.

[0078] (6) Conduct simulation calculations

[0079] After inserting the turbulent data into the grids of the flow field, actual simulation calculations are started.

[0080] (7) Result verification and analysis

[0081] After the simulation calculations are completed, the calculation results are compared and verified with experimental data or known theoretical results to evaluate the accuracy and applicability of the simulation results, and the simulation parameters or the application method of the data are adjusted according to the verification results. If the simulation results are consistent with the actual situation, further flow field analysis and optimization can be carried out to further understand and improve the accuracy and reliability of the flow field simulation.

[0082] As described above, it is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. For those of ordinary skill in the art in this technical field, without departing from the principle described in the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for generating high-precision simulation inflow conditions for reactor thermal fluids, characterized in that: The following steps are involved: S1. Construct a turbulence database, wherein the database includes a data source module, a data storage module, a data reading module and a data application module; S2. Obtain fluid mechanics experimental data, and use RANS, LES and DNS methods to obtain simulation calculation results, and store the obtained experimental data and simulation results in the data source module; S3, dividing the simulation results of the experimental data set obtained in S2 into flow field information files and turbulence data files and storing them in the data storage module, unifying the type and length of the data during storage to improve storage efficiency and reading speed; S4, based on the data reading module, uses multi-threading and parallel I / O technology to read data and improve reading efficiency; S5. After the data is read, before applying the data to the flow field, prepare the network of the flow field to be simulated, set the boundary conditions, and judge whether the turbulence data matches the grid or spatial resolution of the flow field based on the data application module. If they match, directly enter S6; if not, based on the adaptive interpolation algorithm of geometric classification, according to the geometric dimension of the model, the structured and unstructured grid, apply the turbulence data to different flow field models, and then enter S6; The adaptive interpolation algorithm based on geometric classification applies turbulence data to different flow field models according to the geometric dimensions of the model and the structured and unstructured nature of the grid, and specifically includes the following contents: If the model is one-dimensional, it means that the flow field modeling is simple and the data is applied using linear interpolation; If the model is two-dimensional, the network used in the flow field model is divided into structured and unstructured. When the grid constructed by the flow field is a structured grid, the bilinear interpolation method is used to interpolate the physical quantity; if the constructed grid is an unstructured grid, the high-precision interpolation method based on WENO is used for interpolation; If the model is three-dimensional, radial basis function interpolation is used to interpolate turbulence data; S6. Use the input data to perform simulation calculations; S7. After the simulation calculation is completed, the calculation results will be compared with the experimental data or known theoretical results to evaluate the accuracy and applicability of the simulation results, and the simulation parameters or data application methods will be adjusted according to the verification results.

2. A method for generating high-precision simulation inflow conditions for reactor thermal fluid according to claim 1, characterized in that: The flow field information file in S3 is used to store flow field and turbulence information, including turbulence energy spectrum, relevant time scale, turbulence intensity, type of calculation process and flow field information; the turbulence data file is used to store data sets, including point coordinates, velocity component values ​​in various directions and passive scalar values.

3. The method for generating high-precision simulation inflow conditions for reactor thermal fluid according to claim 1, characterized in that: S4 uses multithreading and parallel I / O technology to read data, specifically including the following: S4.1, data segmentation: divide the large file into multiple small blocks, each block contains a certain number of flow field data points; S4.2, multi-threaded reading: Use the standard thread library to create multiple threads, each thread is responsible for reading one or more data blocks; S4.3, data buffering: Each thread stores the read data block into a shared buffer, using a lock mechanism to ensure thread safety; S4.4, Data parsing: The read data block is parsed into specific flow field point data in the buffer, including coordinates, velocity components and passive scalars.

Images