Multi-physical coupling method and system for nuclear reactor model

Abstract

The invention discloses a nuclear reactor model-oriented multi-physical coupling method and system, and relates to the technical field of multi-physical field coupling and advanced nuclear reactors. The system comprises a standardized function interface layer, which is used for splitting each sub-physical field calculation code into independent function modules according to an execution process and packaging the independent function modules into standardized interface functions; the general data structure interaction layer is used for establishing a general multi-dimensional array independent of a specific physical field and a general grid and physical field data structure; the multi-physical field grid mapping and data transfer layer is used for carrying out grid overlapping identification and weight matrix calculation based on a general structure; the process control layer is coupled in a unified mode, sub-modules are dispatched in a coordinated mode through a unified frame, nonlinear iteration, global convergence judgment and asynchronous time step size control are taken over, and a state rollback mechanism is established. According to the method, standardized integration of heterogeneous programs is achieved, and the expansibility, the code reuse rate and the calculation stability of a multi-physics field coupling system are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of multiphysics coupling and advanced nuclear reactor technology, and more specifically, relates to a multiphysics coupling method and system for nuclear reactor models. Background Technology

[0002] A nuclear reactor is a complex system involving the tight coupling of multiple physics fields, including neutron physics, thermal hydraulics, and fuel properties. To accurately predict its safety characteristics under normal operation and accident conditions, high-fidelity multiphysics coupling simulations are essential. Currently, there are two main technical approaches to multiphysics coupling technology: 1) File-based loose coupling: This approach involves calling independent single-physics programs via external scripts and exchanging data using external files. While this method makes full use of existing mature code, frequent file read / write operations lead to low computational efficiency and make it difficult to achieve high-fidelity, tightly coupled computations.

[0003] 2) Tight Coupling Based on Memory: The source code of different physics fields is integrated into the same executable program through hard coding, directly accessing memory addresses for data exchange. This method has high data transfer efficiency and is currently the mainstream coupled data transfer method. However, this method faces serious challenges in practical engineering applications: First, existing physics code often adopts a holistic "input-computation-output" process, with heavy reliance on global variables. Directly modifying it through embedded code not only requires a deep understanding of the original complex logic but also easily compromises code integrity and introduces potential errors. Second, different physics programs lack a unified data interaction standard. To achieve pairwise coupling of N different physics codes, theoretically, N(N-1) dedicated interfaces need to be developed, resulting in extremely poor scalability of the coupled system. Each time a new physics code is introduced, adaptation code must be rewritten for its unique data structure, unit system, and mesh type, causing significant redundant development work. Furthermore, different physics typically employ different spatial discretization methods (such as structured or unstructured meshes), making it difficult to guarantee data conservation in the absence of a standardized mapping mechanism.

[0004] Therefore, there is an urgent need for a method that can perform non-intrusive or low-intrusive modifications to existing code and provide standardized universal interfaces to achieve modularization, efficiency and scalability of multi-physics coupling in nuclear reactors. Summary of the Invention

[0005] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention provides a multi-physics coupling method and system for nuclear reactor models. Its purpose is to solve the technical problems of traditional embedded coupling methods, such as large damage to the original code logic, high risk of modification and difficulty in expansion.

[0006] To achieve the above objectives, according to one aspect of the present invention, a multiphysics coupling system for nuclear reactor models is provided, comprising: A standardized functional interface layer is used to implement non-intrusive modular encapsulation of heterogeneous physics calculation code for nuclear reactors. For heterogeneous sub-physics field codes with different physics fields, the original overall calculation process is broken down into five independent functional modules: initialization, steady-state calculation, transient calculation, data interaction, and termination / release. These functional modules are then encapsulated into a series of standardized interface functions, configured to perform the following operations: firstly, receiving control commands from the unified coupling framework and passing them to the corresponding sub-physics field calculation code; secondly, converting the private data structures within the sub-physics field calculation code into general data structures. This achieves the separation of the internal control logic of the sub-physics field calculation code from the calling logic of the external unified coupling framework. A general data structure interaction layer is used to establish a unified data and mesh description system independent of specific physical fields. Within a unified coupling framework, a general multidimensional array data structure and a general mesh and physical field data structure are constructed. The general multidimensional array data structure serves as the underlying carrier for memory-level data exchange, receiving private data from each sub-physical field transmitted through the interface functions in the standardized functional interface layer. The general mesh and physical field data structure receives private geometric topology information and physical field variables from each sub-physical field transmitted through the interface functions in the standardized functional interface layer. A multiphysics mesh mapping and data transfer layer is used to achieve high-fidelity data transfer between different meshes. Utilizing the data transfer module in the unified coupling framework, it calls the data interaction interfaces of each sub-physics field in the standardized functional interface layer to convert the private data structure of the sub-physics field into the general data structure described in the general data structure interaction layer. Then, based on the general mesh and physics field data structure, mesh overlap identification and weight matrix calculation are performed to realize mesh mapping and data transfer between different physics fields. Finally, the mapped data is converted back from the general data structure to the private data structure of the target sub-physics field, completing the data update in the multiphysics coupling calculation. A unified coupling process control layer is used to realize the coordinated scheduling and nonlinear iterative control of multiphysics coupled systems. Utilizing the overall control module in the unified coupling framework, multiple sub-physics modules are integrated by calling standardized interface functions encapsulated in the standardized function interface layer. A general data structure defined in the general data structure interaction layer serves as the underlying carrier for nonlinear iteration, uniformly managing the strong coupling feedback between multiphysics. During the iteration cycle, the data transfer module described in the multiphysics mesh mapping and data transfer layer achieves high-fidelity mesh mapping and data transfer between different physics. Simultaneously, based on the reactor multiphysics coupling characteristic parameters, it takes over the global convergence judgment of each sub-physics. Based on the time scale differences of the physical processes, it takes over global time step control and triggers a state rollback mechanism when calculation anomalies occur, thereby achieving high-fidelity coupled simulation of the nuclear reactor under normal operation and accident conditions.

[0007] Furthermore, in the standardized functional interface layer, the initialization functional module encapsulates the following interface functions: MPI communication setting interface, data file setting interface, and parameter initialization setting interface; wherein, the MPI communication setting interface is used to receive the MPI communicator passed in by the unified coupling framework to determine the MPI parallel computing environment of the subphysics calculation code in multiphysics; the data file setting interface is used to define the path and name of the input configuration file and output file required by the subphysics calculation code; the parameter initialization setting interface is used to read the input file and perform memory allocation and initialization of the relevant parameters inside the subphysics calculation code.

[0008] Furthermore, in the standardized functional interface layer, the steady-state calculation functional module encapsulates the following interface functions: steady-state preprocessing interface, steady-state complete calculation interface, steady-state single calculation interface, and steady-state postprocessing interface. The steady-state preprocessing function initializes parameters such as the iteration counter and convergence criteria before the steady-state calculation begins. The steady-state complete calculation interface executes the complete steady-state calculation process within the sub-physics calculation code until convergence, preserving its unique iterative solution strategy. The steady-state single calculation interface executes a single steady-state iterative calculation of the sub-physics field, without initialization or post-processing operations, and is cyclically called by the nonlinear iterative solver in the unified coupling framework to achieve unified control of the iterative logic. The steady-state postprocessing interface records and outputs the relevant steady-state calculation results after the steady-state calculation is completed.

[0009] Furthermore, in the standardized functional interface layer, the transient computation functional module encapsulates the following interface functions: transient preparation interface, time step update interface, current time return interface, total time return interface, time step computation preprocessing interface, time step complete computation interface, time step single computation interface, and time step computation postprocessing interface; wherein, the transient preparation interface is used to perform transition settings from steady-state to transient computation; the time step update interface is used to receive the current computation time step specified by the unified coupling framework; the current time return interface is used to return the current computation time of the sub-physics field; and the total time return interface is used to return the current computation time of the sub-physics field. The termination time of the physical field; the time step calculation preprocessing interface is used to perform parameter initialization before the current time step calculation; the time step complete calculation interface is used to execute the complete transient calculation process of the sub-physics calculation code within the current time step until convergence, so as to retain its unique internal iterative solution strategy; the time step single calculation interface is used to execute a single transient iterative calculation of the sub-physics field, without initialization or post-processing operations, and is called cyclically by the nonlinear iterative solver of the unified coupling framework to achieve unified control of the iterative logic; the time step calculation postprocessing interface is used to record and output the relevant transient calculation results after the current time step calculation is completed.

[0010] Furthermore, in the standardized functional interface layer, the data interaction functional module encapsulates the following interface functions: mesh type return interface, mesh coordinate return interface, mesh topology return interface, physical field parameter name return interface, physical field parameter conservation type return interface, physical field parameter storage method return interface, and physical field parameter value return interface; wherein, the mesh type return interface is used to return the mesh type of the sub-physical field; the mesh coordinate return interface is used to return an array of coordinates of all nodes corresponding to the mesh of the sub-physical field in the International System of Units (SI); the mesh topology return interface is used to return an index array of the mesh cell-node connection relationships of the sub-physical field; the physical field parameter name return interface is used to return the names of the physical field parameters that the sub-physical field needs to transfer data with other sub-physical fields; the physical field parameter conservation type return interface is used to return the interpolation conservation property of the physical field parameter; the physical field parameter storage method return interface is used to identify whether the physical field parameter is stored on a mesh node or at the center of a mesh cell; and the physical field parameter value return interface is used to return a one-dimensional array in SI that stores the sub-physical field parameter. The termination and release function module encapsulates the following interface functions: result output interface, memory release interface, and MPI communication termination interface; wherein, the result output interface is used to output the final calculation result; the memory release interface is used to release all dynamically allocated memory; and the MPI communication termination interface is used to release the MPI communicator.

[0011] Furthermore, in the general data structure interaction layer, the general multidimensional array data structure, which is independent of specific physical fields, adopts a layered design, including a bottom-level storage module and a dimension management module. The bottom-level storage module uses a dynamic array with contiguous memory addresses to store the bottom-level data, adopts a generic programming model to support the storage of basic data types and composite constructed data types, and provides memory allocation, release, deep copy, and arithmetic and statistical calculation functions based on operator overloading. On the basis of the contiguous storage of the bottom-level data, the dimension management module encapsulates the multidimensional access interface to the bottom-level data, supports array definitions of different dimensions and slicing operations to extract and copy low-dimensional subarrays from high-dimensional arrays.

[0012] Furthermore, in the general data structure interaction layer, the general mesh and physical field data structure, which are independent of specific physical field calculation codes, adopt a unified node-cell topology description method, which can handle various mesh forms such as hexahedral structured meshes for neutron physics, hexahedral unstructured meshes for thermal-hydraulic fields, and tetrahedral unstructured meshes for fuel performance; it includes an identifier that records the mesh type name, an array that stores the mesh node coordinates, and an array that stores the cell-node connection relationships; and this data structure can store physical field parameters with different physical properties to support the automatic selection of the corresponding interpolation strategy based on the parameter properties during mesh mapping.

[0013] Furthermore, in the multiphysics mesh mapping and data transfer layer, the data transfer module in the unified coupling framework encapsulates the following interface functions: mesh creation interface, mesh mapping initialization interface, physics data transfer interface, physics data return interface, and mesh mapping interface. Specifically, the mesh creation interface calls the mesh class interface functions encapsulated in the standardized function interface layer to convert the private mesh data structure of the sub-physics field into a general mesh data structure. The mesh mapping initialization interface calculates the mesh overlap relationship based on the general mesh topology information of the source and target sub-physics fields, generating and storing a sparse mesh mapping weight matrix. The physics data transfer interface converts the private data structure storing the sub-physics field parameters into a general one-dimensional array in the unified coupling framework and stores it in the general mesh and physics field data structure. The physics data return interface converts the general one-dimensional array stored in the general mesh and physics field data structure back into the private array structure of the sub-physics field. The mesh mapping interface calls the core mapping algorithm to convert the data of the source sub-physics field into a data array on the target sub-physics field mesh according to the mesh mapping weight matrix.

[0014] Furthermore, in the unified coupling process control layer, the overall control module in the unified coupling framework encapsulates the following interface functions: coupling MPI communication initialization interface, coupling initialization interface, solver registration interface, nonlinear iteration interface, coupling steady-state solution interface, global convergence judgment interface, global time step control interface, coupling transient solution interface, state saving interface, state rollback interface, coupling post-processing interface, coupling memory release interface, and coupling MPI communication termination interface; among them, the coupling MPI communication initialization interface is used to initialize the MPI communicator of the coupling program; the coupling initialization interface is used to read the coupling input file and initialize the coupling parameters; the solver registration interface is used to register the encapsulated physics objects to the solver list of the unified coupling framework; the nonlinear iteration interface is used to execute the nonlinear iteration strategy between multiphysics fields, driving each physics field to perform iterative solutions until the convergence criterion is met; the coupling steady-state solution interface is used to execute multiphysics... The main loop of the steady-state coupling achieves unified control over the iteration process of each physics field by calling the functional interfaces in the standardized functional interface layer; the global convergence judgment interface is used to collect the residual variables of each physics field and determine whether the entire coupled system meets the convergence criterion; the global time step control interface is used to allocate the time step of each sub-physics field; the coupled transient solution interface is used to execute the main loop of multi-physics transient coupling, achieving unified control over the iteration process of each physics field by calling the functional interfaces in the standardized functional interface layer; the state saving interface is used to save global state parameters such as the number of coupling iteration steps and time; the state rollback interface is used to control all physics fields to roll back to the last successful time step and adjust the calculation strategy to recalculate when the physics field calculation is abnormal; the coupling post-processing interface is used to record and output the simulation results in the coupled framework; the coupling memory release interface is used to release all dynamically allocated memory in the coupled framework; and the coupling MPI communication termination interface is used to release the MPI communicator.

[0015] According to another aspect of the present invention, a multi-physics coupling method for nuclear reactor models is provided, implemented based on the multi-physics coupling system as described in any of the preceding claims, comprising: First, the coupled MPI initialization interface is called to establish a global parallel communication environment; Subsequently, coupling parameter initialization is performed, and the global configuration file is read to determine key control parameters; Next, each sub-physics field registers its encapsulated physics field object to the solver list of the unified coupling framework through the solver registration interface, thus completing the construction of the global environment; After the global environment is ready, the initialization process of each physical field is triggered: each sub-physical field calls the MPI communication setting interface to establish an internal communication subdomain, reads the mesh, boundary conditions and material properties through the data file setting interface, and completes the initial state setting of its respective solver through the parameter initialization setting interface. Subsequently, according to the preset strategy, the program enters the steady-state or transient calculation branch, and the main control program sequentially calls the sub-physics solver interface. Among them, the output data of the upstream physics field is mapped to the coupling parameters required by the downstream physics field through the mesh mapping module. Then, the multi-physics coupling iteration is promoted by the nonlinear iteration interface, and the global physics field residual is checked after each iteration. If the global convergence judgment criterion is not met, the program returns to the calculation entry point to continue the iteration. If convergence is achieved, the program enters the next time step or ends the calculation. After a single-step calculation is completed, a state save / state rollback operation is performed; if the calculation diverges or is abnormal, the state rollback mechanism is used to restore to the previous convergence point and adjust the step size to iterate again. After the coupled calculation is completed, the parallel execution results output interface and memory release interface of each sub-physics field are provided. Finally, the framework executes the coupled post-processing interface to collect global data, calls the coupled memory release interface to clean up shared resources, and shuts down the parallel environment through the coupled MPI communication termination interface.

[0016] In summary, the technical solutions conceived in this invention, compared with the prior art, can achieve the following beneficial effects: 1. This invention decouples the physical calculation core of the subphysics code from the data communication logic, thereby achieving standardized integration of heterogeneous programs and significantly improving the scalability, code reusability, and computational stability of multiphysics coupling systems.

[0017] 2. By splitting the code and encapsulating the interfaces, the physical computing core and data communication logic were decoupled. This design allows the original monolithic code to be called by modern coupled frameworks in a modular form, greatly extending the code's lifecycle and reducing maintenance costs.

[0018] 3. The unified interface design and general data structure based on the unified coupling framework overcome the data transfer problem between different physics field codes, avoid the repetitive work of writing dedicated interfaces for multiple physics field codes, and greatly improve the scalability of the system.

[0019] 4. The proposed general multidimensional array data structure adopts a two-layer data structure design, realizing efficient multidimensional data management and operation. By providing rich interfaces and operation methods, users can easily access data, perform calculations, and extract subarrays, significantly reducing the development difficulty and memory management risks when processing large-scale multiphysics data.

[0020] 5. By adopting a universal grid and physical field data structure, the correct interpolation algorithm can be matched according to the physical properties of the transmitted parameters when performing grid mapping between different grids, thus strictly guaranteeing the accuracy of data transmission.

[0021] 6. A unified coupling framework is used to take over global time step control and nonlinear iterative processes. This not only effectively solves the synchronization problem caused by the huge differences in time scales between different physical fields, but more importantly, through a unified global convergence judgment and anomaly backoff mechanism, it significantly suppresses the numerical oscillation phenomenon common in strongly coupled calculations, and greatly improves the convergence stability and computational accuracy of nonlinear coupled systems. Attached Figure Description

[0022] Figure 1 This is a general framework diagram of a multiphysics field unified coupling system for nuclear reactors according to a preferred embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of a multiphysics coupling system for nuclear reactors according to a preferred embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of a 2×2 small core coupling example of a preferred embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the average coolant temperature distribution in a preferred embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0027] like Figure 1 As shown, this embodiment provides a multiphysics coupling system for nuclear reactor models, specifically including: A standardized functional interface layer is constructed to achieve non-intrusive modular encapsulation of heterogeneous physics calculation code for nuclear reactors. For heterogeneous sub-physics field codes with different physical fields (such as neutron transport / diffusion, thermal-hydraulic dynamics, and fuel performance), the original overall calculation process is broken down into five independent functional modules: "initialization," "steady-state calculation," "transient calculation," "data interaction," and "end release." These functional modules are encapsulated into a series of standardized interface functions, configured to perform the following operations: First, receiving control commands from the unified coupling framework and passing them to the corresponding sub-physics field calculation code. Second, converting the private data structures within the sub-physics field calculation code into general data structures. This separates the internal control logic of the sub-physics field calculation code from the calling logic of the external unified coupling framework.

[0028] A general data structure interaction layer is constructed to establish a unified data and grid description system independent of specific physical fields. Within a unified coupling framework, a general multidimensional array data structure and a general grid and physical field data structure are built. The general multidimensional array serves as the underlying carrier for memory-level data exchange, used to receive private data from various sub-physics fields (such as full-core power distribution, full-core fuel temperature distribution, and core coolant distribution) transmitted from the interface functions of the standardized functional interface layer. The general grid and physical field data structure is used to receive private geometric topology information and physical field variables from various sub-physics fields (such as structured grids for neutron physics / diffusion, and unstructured grids for fuel performance) transmitted from the interface functions of the standardized functional interface layer.

[0029] A multiphysics grid mapping and data transfer layer is constructed to achieve high-fidelity data transfer between different grids. Utilizing the data transfer module within the unified coupling framework, the data interaction interfaces of each subphysics field described in the standardized functional interface layer are invoked to convert the private data structures of the subphysics fields into a general data structure as described in the general data structure interaction layer. Then, based on the general grid and physics field data structures, grid overlap identification and weight matrix calculation are performed to achieve grid mapping and data transfer between different physics fields (e.g., the conservation mapping of neutron-generated thermal power between the neutron physics grid and the thermal-hydraulic grid). Finally, the mapped data is converted back from the general data structure to the private data structure of the target subphysics field, completing the data update in the multiphysics coupling calculation.

[0030] A unified coupled process control layer is constructed to achieve coordinated scheduling and nonlinear iterative control of the multiphysics coupled system. Utilizing the overall control module within the unified coupling framework, multiple sub-physics modules are integrated by calling standardized interface functions encapsulated in the standardized function interface layer. A general data structure defined in the general data structure interaction layer serves as the underlying carrier for nonlinear iteration, uniformly managing strong coupling feedbacks between multiphysics fields (such as Doppler effect feedback and coolant density feedback). During the iteration cycle, the data transfer module described in the multiphysics mesh mapping and data transfer layer achieves high-fidelity mesh mapping and data transfer between different physics fields. Simultaneously, based on the reactor multiphysics coupling characteristic parameters (such as the effective multiplication coefficient, total core power, and node peak temperature), the system takes over the global convergence judgment of each sub-physics field. Based on the time scale differences of the physical processes (such as microsecond-level changes in neutronics and second-level changes in thermal hydraulics), the system takes over global time step control and triggers a state rollback mechanism when calculation anomalies occur, thereby achieving high-fidelity coupled simulation of the nuclear reactor under normal operation and accident conditions.

[0031] The method in this embodiment is mainly used for multiphysics coupling simulation of nuclear reactors. It can perform non-intrusive simulation of existing code and provides a standardized, universal interface to achieve modularity, efficiency, and scalability of multiphysics coupling in nuclear reactors. The specific steps and implementation process are as follows: In a specific embodiment, the "initialization" function module interface of the standardized functional interface layer is encapsulated as follows: MPI communication setting interface, data file setting interface, and parameter initialization setting interface. The MPI communication setting interface receives the MPI communicator from the unified coupling framework to determine the MPI parallel computing environment for the subphysics calculation code in the multiphysics model. The data file setting interface defines the paths and names of the input configuration files and output files required by the subphysics calculation code. The parameter initialization setting interface reads the input file and performs memory allocation and initialization for relevant parameters within the subphysics calculation code (such as reactor geometry parameters, material cross-sections, and initial control rod steps required for the neutron physics field; and reactor geometry parameters, inlet initial pressure, and flow distribution required for the thermal-hydraulic field).

[0032] In a specific embodiment, the "steady-state calculation" function module interface of the standardized functional interface layer is encapsulated as follows: steady-state preprocessing interface, steady-state complete calculation interface, steady-state single-step calculation interface, and steady-state post-processing interface. The steady-state preprocessing function initializes parameters such as the iteration counter and convergence criteria before the steady-state calculation begins. The steady-state complete calculation interface executes the complete steady-state calculation process within the sub-physics calculation code until convergence, preserving its unique iterative solution strategy. The steady-state single-step calculation interface executes a single steady-state iterative calculation of the sub-physics field. This interface does not contain any initialization or post-processing operations and can be repeatedly called by the nonlinear iterative solver in the unified coupling framework to achieve unified control of the iterative logic. The steady-state post-processing interface records and outputs the relevant steady-state calculation results after the steady-state calculation is completed.

[0033] In a specific embodiment, the "transient computation" function module interface of the standardized functional interface layer is encapsulated as follows: transient preparation interface, time step update interface, current time return interface, total time return interface, time step computation preprocessing interface, time step complete computation interface, time step single computation interface, and time step computation postprocessing interface. Specifically, the transient preparation interface is used to perform transition settings from steady-state to transient computation. The time step update interface is used to receive the current computation time step specified by the unified coupling framework. The current time return interface is used to return the current computation time of the sub-physics field. The total time return interface is used to return the termination time of the sub-physics field. The time step computation preprocessing interface is used to perform parameter initialization before the current time step computation. The time step complete computation interface is used to execute the complete transient computation process of the sub-physics computation code within the current time step until convergence, to preserve its unique internal iterative solution strategy. The time step single computation interface is used to execute a single transient iterative computation of the sub-physics field. This interface also does not contain initialization or post-processing operations and can be repeatedly called by the nonlinear iterative solver of the unified coupling framework to achieve unified control of the iterative logic. The time step calculation post-processing interface is used to record and output the relevant transient calculation results after the current time step calculation is completed.

[0034] In a specific embodiment, the "data interaction" function module interface of the standardized functional interface layer is encapsulated as follows: Mesh type return interface, mesh coordinate return interface, mesh topology return interface, physics parameter name return interface, physics parameter conservation type return interface, physics parameter storage method return interface, and physics parameter value return interface. Specifically, the mesh type return interface returns the mesh type of the sub-physics field (e.g., hexahedral mesh, tetrahedral mesh, and triangular prism mesh). The mesh coordinate return interface returns an array of coordinates of all nodes conforming to the International System of Units (SI) corresponding to the sub-physics field mesh. The mesh topology return interface returns an index array of the mesh cell-node connection relationships of the sub-physics field. The physics parameter name return interface returns the names of physics parameters that the sub-physics field needs to transfer data with other sub-physics fields (e.g., relative power density output by the neutron physics field, coolant temperature output by the thermal-hydraulic field, fuel center temperature output by the fuel performance field, etc.). The physics parameter conservation type return interface returns the interpolation conservation properties of the physics parameters (e.g., intensity conservation, intensity maximization, extension conservation, and extension maximization). The physics parameter storage method return interface is used to identify whether the physics parameters are stored on grid nodes or at the center of grid cells. The physics parameter value return interface is used to return a one-dimensional array in SI units that stores the sub-physics parameters.

[0035] In a specific embodiment, the "End Release" function module interface of the standardized functional interface layer is encapsulated as follows: a result output interface, a memory release interface, and an MPI communication termination interface. The result output interface is used to output the final calculation result. The memory release interface is used to release all dynamically allocated memory. The MPI communication termination interface is used to release the MPI communicator.

[0036] In a specific embodiment, the general multidimensional array data structure of the general data structure interaction layer adopts a layered design, including a bottom-level storage module and a dimension management module. The bottom-level storage module uses a dynamic array with contiguous memory addresses to store the bottom-level data. It employs a generic programming model to support the storage of basic data types and composite constructed data types, and provides memory allocation, release, deep copy, and arithmetic operations (addition, subtraction, multiplication, and division) and statistical calculations (maximum, minimum, and L2 norm, etc.) based on operator overloading. The dimension management module, based on the contiguous storage of the bottom-level data, encapsulates a multidimensional access interface for the bottom-level data, supporting array definitions of arbitrary dimensions and slicing operations to extract and copy low-dimensional subarrays from high-dimensional arrays.

[0037] In one specific embodiment, the general mesh and physical field data structure of the general data structure interaction layer adopt a unified "node-cell" topology description method, which can handle various mesh forms such as hexahedral structured meshes for neutron physics, hexahedral unstructured meshes for thermal-hydraulic fields, and tetrahedral unstructured meshes for fuel performance. It includes an identifier recording the mesh type name, an array storing mesh node coordinates, and an array storing cell-node connection relationships. Furthermore, this data structure can store physical field parameters with different physical properties (such as intensity or extension) to support automatic selection of appropriate interpolation strategies based on parameter properties during mesh mapping.

[0038] In a specific embodiment, the data transfer module of the multiphysics mesh mapping and data transfer layer encapsulates the following interface functions: mesh creation interface, mesh mapping initialization interface, physics data transfer interface, physics data return interface, and mesh mapping interface. Specifically, the mesh creation interface calls the mesh class interface functions encapsulated in the standardized functional interface layer to convert the private mesh data structure of the sub-physics field into a general mesh data structure. The mesh mapping initialization interface calculates the mesh overlap relationship based on the general mesh topology information of the source and target sub-physics fields, generating and storing a sparse mesh mapping weight matrix. The physics data transfer interface converts the private data structure storing the sub-physics field parameters into a general one-dimensional array within the unified coupling framework and stores it in the general mesh and physics field data structure. The physics data return interface converts the general one-dimensional array stored in the general mesh and physics field data structure back into the private array structure of the sub-physics field. The mesh mapping interface calls the core mapping algorithm to convert the data of the source sub-physics field into a data array on the target sub-physics field mesh according to the mesh mapping weight matrix.

[0039] In a specific embodiment, the overall control module of the unified coupling process control layer encapsulates the following interface functions: coupling MPI communication initialization interface, coupling initialization interface, solver registration interface, nonlinear iteration interface, coupling steady-state solution interface, global convergence judgment interface, global time step control interface, coupling transient solution interface, state saving interface, state rollback interface, coupling post-processing interface, coupling memory release interface, and coupling MPI communication termination interface. Specifically, the coupling MPI communication initialization interface initializes the MPI communicator of the coupling program. The coupling initialization interface reads the coupling input file and initializes the coupling parameters. The solver registration interface registers the encapsulated physics objects to the solver list of the unified coupling framework. The nonlinear iteration interface executes nonlinear iteration strategies between multiple physics (such as operator splitting, fixed-point iteration, or Newton-Krylov algorithm without Jacobian matrix), driving each physics to iteratively solve until the convergence criterion is met. The coupling steady-state solution interface executes the main loop of multi-physics steady-state coupling, achieving unified control of the iteration process of each physics by calling the function interfaces in the standardized function interface layer. The global convergence judgment interface collects the residual variables of each physics field to determine whether the entire coupled system meets the convergence criteria. The global time step control interface allocates the time step size for each sub-physics field. The coupled transient solution interface executes the main loop for multi-physics transient coupling, achieving unified control of the iteration process of each physics field by calling functional interfaces in the standardized functional interface layer. The state saving interface saves global state parameters such as the number of coupling iteration steps and time. The state rollback interface controls all physics fields to roll back to the last successful time step and adjust the calculation strategy for recalculation when a physics calculation fails. The coupling post-processing interface records and outputs the simulation results in the coupled framework. The coupling memory release interface releases dynamically allocated memory in all coupled frameworks. The coupled MPI communication termination interface releases the MPI communicator.

[0040] like Figure 1The unified coupling process control layer shown follows a typical lifecycle management pattern of "initialization - core iterative calculation - state management - resource release". First, the coupling MPI initialization interface is called to establish a global parallel communication environment, ensuring smooth message transmission between solvers of each physics field. Then, coupling parameter initialization is performed, reading the global configuration file to determine key control parameters such as time step, coupling algorithm, and convergence criterion. Next, each sub-physics field registers its encapsulated physics field object to the solver list of the unified coupling framework through the solver registration interface, completing the construction of the global environment. After the global environment is ready, the architecture triggers the initialization process of each physics field. Each sub-physics field calls the MPI communication setting interface to establish an internal communication subdomain, reads the mesh, boundary conditions, and material properties through the data file setting interface, and completes the initial state setting of its respective solver through the parameter initialization setting interface. Subsequently, the framework enters the steady-state or transient calculation branch according to a preset strategy, and the main control program sequentially calls the sub-physics field solver interfaces. Among these, the mesh mapping module maps the output data of the upstream physics field to the coupling parameters required by the downstream physics field. Next, the multiphysics coupling iteration is advanced using a nonlinear iterative interface, and the global physics residual is checked after each iteration. If the global convergence criterion is not met, the iteration returns to the computation entry point and continues; if convergence is achieved, the iteration proceeds to the next time step or terminates. After a single-step computation, a state save / state rollback operation is performed. If the computation diverges or malfunctions, the state rollback mechanism restores the system to the previous convergence point and adjusts the step size for a new iteration. After the coupling computation is completed, the result output interface and memory release interface of each subphysics field are executed in parallel. Finally, the framework executes the coupling post-processing interface to collect global data, calls the coupling memory release interface to clean up shared resources, and closes the parallel environment through the coupling MPI communication termination interface.

[0041] like Figure 2 As shown, taking the coupled simulation of neutron physics and thermal-hydraulic processes in a nuclear reactor core as an example, this paper elaborates on the control logic and data interaction mechanism of the unified coupling framework for the multi-physics coupling process. First, the system calls the coupling MPI communication initialization interface to build a parallel environment and loads global control parameters through the coupling initialization interface. Then, it enters the sub-physics interface initialization process, where the neutron physics and thermal-hydraulic field codes perform the following operations in parallel: First, the solver registration interface is called to register its solver information with the framework. Next, the mesh creation interface is called to generate the computational mesh. Based on the generated mesh, a general mesh and physics data structure are constructed, standardizing the heterogeneous physics data and laying the foundation for subsequent data exchange. After initialization, the system selects either steady-state or transient calculation mode based on the control parameters.

[0042] If the system enters the steady-state calculation branch, it first executes the steady-state preprocessing interface, followed by the steady-state coupled iterative loop: First, it calls the neutron physics steady-state calculation interface to solve for neutron flux and power distribution. Data is passed to the framework via the neutron physics data transfer interface, processed by the general multidimensional array storage and mesh mapping interface, and then the power distribution is passed to the thermal-hydraulic field via the thermal-hydraulic field data feedback interface. Using the updated power data, the thermal-hydraulic field steady-state calculation interface is called to solve for the temperature and flow fields. Then, a reverse mapping is performed, passing feedback data such as temperature and density back to the neutron physics field via the neutron physics data feedback interface through the thermal-hydraulic field data transfer interface, the general multidimensional array, and the mesh mapping interface. At the end of each iteration, the state saving interface is called. The system first checks for steady-state calculation anomalies; if an anomaly is found, the state rollback interface is triggered; otherwise, the steady-state convergence status is checked. If convergence fails, the system returns to the neutron physics steady-state calculation interface to restart the loop; if convergence occurs, the steady-state post-processing interface is executed.

[0043] If transient calculations are initiated (after the steady-state calculations are completed), the transient preparation interface is first called, and a time-stepping loop is entered under global time-step control: the neutron physics field time-step calculation interface and the thermal-hydraulic field time-step calculation interface are called sequentially, and a bidirectional mesh mapping operation is performed between them (same as the data transfer logic in the steady-state phase). At the end of each step, the state saving interface is called. If a time-step calculation anomaly is detected, the state rollback interface is triggered, and the time-step may be adjusted; if normal and the time-step convergence condition is met, the time-step calculation post-processing interface is executed. It is checked whether the termination time has been reached; if not, the next time-step loop begins; otherwise, the transient calculation ends.

[0044] After all computational tasks are completed, the process enters the resource release phase. First, the sub-physics post-processing interface is called, and each physics executes its result output interface to write data to disk and then calls the memory release interface to destroy local objects. Next, the coupled post-processing interface is executed to perform global data statistics and output. Finally, the coupled memory release interface is called to clean up shared framework resources, and the parallel environment is shut down via the coupled MPI communication termination interface.

[0045] To verify the effectiveness of the method of this invention, a simplified model based on the PWR / MOX benchmark problem was constructed as follows. Figure 3The 2×2 small core coupling example is shown. The proposed multiphysics coupling method was applied to this example to test its performance. The physics program uses 140 grids axially, 20 grids each for the upper and lower reflectors, with each grid size of 2.142 cm, and 100 grids for the core active region, with each grid size of 3.6576 cm. The subchannel program only models the core active region axially, using 25 grids. Except for the last grid, all grid sizes are set to 15 cm, with the last grid size set to 5.56 cm. The significant difference in grid sizes between the two methods tested the coupling capability between mismatched grids. The calculation results are shown in Table 1. The effective multiplication coefficient calculated by this method deviates from the reference solution by 158 pcm (0.00158), the average Doppler temperature deviates from the reference solution by 3.16 °C, and the average fuel rod center temperature deviates from the reference solution by 9.29 °C. The average coolant temperature distribution is shown in Table 1. Figure 4 As shown, compared with the reference solution, the average deviation is 2.33℃, and the maximum deviation is 9.66℃. These results demonstrate that the proposed method successfully achieves modular coupling of heterogeneous code while ensuring computational accuracy, meeting the requirements of high-fidelity simulation in engineering applications.

[0046] Table 1 Comparison of this program and the reference solution in the embodiments

[0047] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements 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 multiphysics coupling system for nuclear reactor models, characterized in that, include: A standardized functional interface layer is used to implement non-intrusive modular encapsulation of heterogeneous physics calculation code for nuclear reactors. For heterogeneous sub-physics field codes with different physics fields, the original overall calculation process is broken down into five independent functional modules: initialization, steady-state calculation, transient calculation, data interaction, and termination / release. These functional modules are then encapsulated into a series of standardized interface functions, configured to perform the following operations: firstly, receiving control commands from the unified coupling framework and passing them to the corresponding sub-physics field calculation code; secondly, converting the private data structures within the sub-physics field calculation code into general data structures. This achieves the separation of the internal control logic of the sub-physics field calculation code from the calling logic of the external unified coupling framework. A general data structure interaction layer is used to establish a unified data and mesh description system independent of specific physical fields. Within a unified coupling framework, a general multidimensional array data structure and a general mesh and physical field data structure are constructed. The general multidimensional array data structure serves as the underlying carrier for memory-level data exchange, receiving private data from each sub-physical field transmitted through the interface functions in the standardized functional interface layer. The general mesh and physical field data structure receives private geometric topology information and physical field variables from each sub-physical field transmitted through the interface functions in the standardized functional interface layer. A multiphysics mesh mapping and data transfer layer is used to achieve high-fidelity data transfer between different meshes. Utilizing the data transfer module in the unified coupling framework, it calls the data interaction interfaces of each sub-physics field in the standardized functional interface layer to convert the private data structure of the sub-physics field into the general data structure described in the general data structure interaction layer. Then, based on the general mesh and physics field data structure, mesh overlap identification and weight matrix calculation are performed to realize mesh mapping and data transfer between different physics fields. Finally, the mapped data is converted back from the general data structure to the private data structure of the target sub-physics field, completing the data update in the multiphysics coupling calculation. A unified coupling process control layer is used to realize the coordinated scheduling and nonlinear iterative control of multi-physics coupling systems. By using the overall control module in the unified coupling framework, multiple sub-physics modules are integrated by calling the standardized interface functions encapsulated in the standardized function interface layer. The system utilizes a general data structure defined in the interaction layer as the underlying carrier for nonlinear iteration, uniformly managing the strong coupling feedback between multiple physics fields. During the iteration cycle, the system uses the data transfer module described in the multiphysics field mesh mapping and data transfer layer to achieve high-fidelity mesh mapping and data transfer between different physics fields. Simultaneously, it takes over the global convergence judgment of each sub-physics field based on the reactor multiphysics field coupling characteristic parameters. Based on the time scale differences of the physical process, it takes over the global time step control and triggers the state rollback mechanism when calculation anomalies occur, thereby realizing high-fidelity coupled simulation of the nuclear reactor under normal operation and accident conditions.

2. The multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the standardized functional interface layer, the initialization functional module encapsulates the following interface functions: MPI communication setting interface, data file setting interface, and parameter initialization setting interface. The MPI communication setting interface receives the MPI communicator from the unified coupling framework to determine the MPI parallel computing environment for the subphysics computation code in multiphysics. The data file setting interface defines the paths and names of the input configuration files and output files required by the subphysics computation code. The parameter initialization setting interface reads the input file and performs memory allocation and initialization for the relevant parameters within the subphysics computation code.

3. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the standardized functional interface layer, the steady-state calculation functional module encapsulates the following interface functions: steady-state preprocessing interface, steady-state complete calculation interface, steady-state single calculation interface, and steady-state postprocessing interface. The steady-state preprocessing function initializes parameters such as the iteration counter and convergence criteria before the steady-state calculation begins. The steady-state complete calculation interface executes the complete steady-state calculation process within the sub-physics calculation code until convergence, preserving its unique iterative solution strategy. The steady-state single calculation interface executes a single steady-state iterative calculation of the sub-physics field, without initialization or post-processing operations, and is cyclically called by the nonlinear iterative solver in the unified coupling framework to achieve unified control of the iterative logic. The steady-state postprocessing interface records and outputs the relevant steady-state calculation results after the steady-state calculation is completed.

4. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the standardized functional interface layer, the transient computation functional module encapsulates the following interface functions: transient preparation interface, time step update interface, current time return interface, total time return interface, time step computation preprocessing interface, time step complete computation interface, time step single computation interface, and time step computation postprocessing interface; wherein, the transient preparation interface is used to perform transition settings from steady-state to transient computation; the time step update interface is used to receive the current computation time step specified by the unified coupling framework; the current time return interface is used to return the current computation time of the sub-physics field; and the total time return interface is used to return the total time of the sub-physics field. Termination time; the time step calculation preprocessing interface is used to perform parameter initialization before the current time step calculation; the time step complete calculation interface is used to execute the complete transient calculation process of the sub-physics calculation code within the current time step until convergence, so as to retain its unique internal iterative solution strategy; the time step single calculation interface is used to execute a single transient iterative calculation of the sub-physics field, without initialization or post-processing operations, and is called cyclically by the nonlinear iterative solver of the unified coupling framework to achieve unified control of the iterative logic; the time step calculation postprocessing interface is used to record and output the relevant transient calculation results after the current time step calculation is completed.

5. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the standardized functional interface layer, the data interaction functional module encapsulates the following interface functions: mesh type return interface, mesh coordinate return interface, mesh topology return interface, physics parameter name return interface, physics parameter conservation type return interface, physics parameter storage method return interface, and physics parameter value return interface. Specifically, the mesh type return interface returns the mesh type of the sub-physics field; the mesh coordinate return interface returns an array of coordinates of all nodes corresponding to the sub-physics field mesh in the International System of Units (SI); the mesh topology return interface returns an index array of the mesh cell-node connection relationships of the sub-physics field; the physics parameter name return interface returns the names of the physics parameters that the sub-physics field needs to transfer data to other sub-physics fields; the physics parameter conservation type return interface returns the interpolation conservation property of the physics parameters; the physics parameter storage method return interface identifies whether the physics parameters are stored on mesh nodes or at the center of mesh cells; and the physics parameter value return interface returns a one-dimensional array in SI that stores the sub-physics field parameters. The termination and release function module encapsulates the following interface functions: result output interface, memory release interface, and MPI communication termination interface; wherein, the result output interface is used to output the final calculation result; the memory release interface is used to release all dynamically allocated memory; and the MPI communication termination interface is used to release the MPI communicator.

6. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the general data structure interaction layer, the general multidimensional array data structure, which is independent of specific physical fields, adopts a layered design, including a bottom-level storage module and a dimension management module. The bottom-level storage module uses a dynamic array with contiguous memory addresses to store the bottom-level data. It adopts a generic programming model to support the storage of basic data types and composite constructed data types, and provides memory allocation, release, deep copy, and arithmetic and statistical calculation functions based on operator overloading. On the basis of the contiguous storage of the bottom-level data, the dimension management module encapsulates the multidimensional access interface to the bottom-level data, supports the definition of arrays of different dimensions, and the slicing operation of extracting and copying low-dimensional subarrays from high-dimensional arrays.

7. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the general data structure interaction layer, the general mesh and physical field data structure, which are independent of specific physical field calculation codes, adopt a unified node-cell topology description method. It can handle various mesh forms such as hexahedral structured meshes for neutron physics, hexahedral unstructured meshes for thermal-hydraulic fields, and tetrahedral unstructured meshes for fuel performance. It includes an identifier that records the mesh type name, an array that stores the mesh node coordinates, and an array that stores the cell-node connection relationships. Furthermore, this data structure can store physical field parameters with different physical properties to support the automatic selection of the appropriate interpolation strategy based on the parameter properties during mesh mapping.

8. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the multiphysics mesh mapping and data transfer layer, the data transfer module in the unified coupling framework encapsulates the following interface functions: mesh creation interface, mesh mapping initialization interface, physics data transfer interface, physics data return interface, and mesh mapping interface. Specifically, the mesh creation interface calls the mesh class interface functions encapsulated in the standardized function interface layer to convert the private mesh data structure of the sub-physics field into a general mesh data structure. The mesh mapping initialization interface calculates the mesh overlap relationship based on the general mesh topology information of the source and target sub-physics fields, generating and storing a sparse mesh mapping weight matrix. The physics data transfer interface converts the private data structure storing the sub-physics field parameters into a general one-dimensional array in the unified coupling framework and stores it in the general mesh and physics field data structure. The physics data return interface converts the general one-dimensional array stored in the general mesh and physics field data structure back into the private array structure of the sub-physics field. The mesh mapping interface calls the core mapping algorithm to convert the data of the source sub-physics field into a data array on the target sub-physics field mesh according to the mesh mapping weight matrix.

9. A multiphysics coupling system for nuclear reactor models as described in claim 1, characterized in that, In the unified coupling process control layer, the overall control module of the unified coupling framework encapsulates the following interface functions: coupling MPI communication initialization interface, coupling initialization interface, solver registration interface, nonlinear iteration interface, coupling steady-state solution interface, global convergence judgment interface, global time step control interface, coupling transient solution interface, state saving interface, state rollback interface, coupling post-processing interface, coupling memory release interface, and coupling MPI communication termination interface. Specifically, the coupling MPI communication initialization interface is used to initialize the MPI communicator of the coupling program; the coupling initialization interface is used to read the coupling input file and initialize the coupling parameters; the solver registration interface is used to register the encapsulated physics objects to the solver list of the unified coupling framework; the nonlinear iteration interface is used to execute the nonlinear iteration strategy between multiple physics fields, driving each physics field to perform iterative solutions until the convergence criterion is met; and the coupling steady-state solution interface is used to execute the multi-physics steady-state solution. The main loop of the multi-physics coupling system calls functional interfaces in the standardized functional interface layer to achieve unified control over the iteration process of each physics field. The global convergence judgment interface collects the residual variables of each physics field and determines whether the entire coupled system meets the convergence criterion. The global time step control interface allocates the time step of each sub-physics field. The coupled transient solution interface executes the main loop of multi-physics transient coupling and achieves unified control over the iteration process of each physics field by calling functional interfaces in the standardized functional interface layer. The state saving interface saves global state parameters such as the number of coupled iteration steps and time. The state rollback interface controls all physics fields to roll back to the last successful time step and adjusts the calculation strategy to recalculate when the physics calculation is abnormal. The coupling post-processing interface records and outputs the simulation results in the coupled framework. The coupling memory release interface releases all dynamically allocated memory in the coupled framework. The coupling MPI communication termination interface releases the MPI communicator.

10. A multiphysics coupling method for nuclear reactor models, implemented based on the multiphysics coupling system described in any one of claims 1 to 9, characterized in that, include: First, the coupled MPI initialization interface is called to establish a global parallel communication environment; Subsequently, coupling parameter initialization is performed, and the global configuration file is read to determine key control parameters; Next, each sub-physics field registers its encapsulated physics field object to the solver list of the unified coupling framework through the solver registration interface, thus completing the construction of the global environment; After the global environment is ready, the initialization process of each physical field is triggered: each sub-physical field calls the MPI communication setting interface to establish an internal communication subdomain, reads the mesh, boundary conditions and material properties through the data file setting interface, and completes the initial state setting of its respective solver through the parameter initialization setting interface. Subsequently, according to the preset strategy, the program enters the steady-state or transient calculation branch, and the main control program sequentially calls the sub-physics solver interface. Among them, the output data of the upstream physics field is mapped to the coupling parameters required by the downstream physics field through the mesh mapping module. Then, the multi-physics coupling iteration is promoted by the nonlinear iteration interface, and the global physics field residual is checked after each iteration. If the global convergence judgment criterion is not met, the program returns to the calculation entry point to continue the iteration. If convergence is achieved, the program enters the next time step or ends the calculation. After a single-step calculation is completed, a state save / state rollback operation is performed; if the calculation diverges or is abnormal, the state rollback mechanism is used to restore to the previous convergence point and adjust the step size to iterate again. After the coupled calculation is completed, the parallel execution results output interface and memory release interface of each sub-physics field are provided. Finally, the framework executes the coupled post-processing interface to collect global data, calls the coupled memory release interface to clean up shared resources, and shuts down the parallel environment through the coupled MPI communication termination interface.

Images