Numerical nuclear reactor system based on supercomputing

Abstract

The invention provides a supercomputing-based numerical nuclear reactor system. The system comprises a user interface, a multi-physical simulation environment, a support tool library, a simulation database and a hardware support environment, the user interface is used for receiving parameters and simulation requirements required by nuclear reactor simulation; the multi-physical simulation environment comprises a multi-physical core calculation part and a coupler and is used for realizing nuclear reactor core physical process calculation and coupling calculation; the support tool library provides support for calculation of each physical process; the simulation database is used for storing simulation data; the hardware supporting environment is used for providing hardware resources. According to the method, the accuracy and credibility of a simulation result are ensured.

Description

Technical Field

[0001] The present invention belongs to the cross - field of high - performance computing and nuclear engineering, and particularly relates to a numerical nuclear reactor system based on supercomputing. Background Art

[0002] Nuclear reactor engineering technology has long faced problems such as large - scale experiments with high difficulty, high cost, and long cycle, making it very difficult to fully verify the rationality and safety of reactor designs through experiments alone. Numerical nuclear reactors (referred to as numerical reactors for short) based on advanced coupled modeling and large - scale parallel computing technology have become a necessary way to solve the above problems. Large - scale, high - fidelity simulations of various computational processes in numerical reactors (such as neutron transport, thermal - hydraulic, structural mechanics, fuel behavior, etc.), as well as multi - scale, strongly non - linear couplings, are quite complex, and the requirements for computational volume and storage are extremely large, requiring the powerful computing power and storage capacity of supercomputing computers (referred to as supercomputers for short). Currently, all supercomputing computers adopt a heterogeneous hybrid architecture of "CPU + accelerator", and more than 90% of the computing power comes from accelerator cores. Therefore, relying on the establishment of an advanced numerical nuclear reactor system with a heterogeneous architecture is of great significance for achieving the autonomy and controllability of reactor technology.

[0003] However, due to the complexity of the reactor system, most existing numerical reactor systems remain at the conceptual design stage and have deficiencies in terms of functional integrity, integration level, simulation fidelity, scale, and accuracy, and cannot be applied in practice. Summary of the Invention

[0004] In order to solve the problems existing in the above - mentioned existing numerical reactor technologies and systems, the present invention provides a numerical nuclear reactor system based on supercomputing.

[0005] The technical solution of the present invention is as follows:

[0006] A numerical nuclear reactor system based on supercomputing, the system includes a user interface, a multi - physical simulation environment, a support tool library and a simulation database, and a hardware support environment; the user interface is used to receive the parameters and simulation requirements required for nuclear reactor simulation; the multi - physical simulation environment includes multi - physical core computing components and a coupler, and is used to implement the core physical process calculation and coupling calculation of the nuclear reactor; the support tool library provides support for the calculation of each physical process; the simulation database is used to store simulation data; the hardware support environment is used to provide hardware resources.

[0007] Furthermore, the multi-physics core computing component includes: a neutron transport computing component, a thermal-hydraulic computing component, a structural mechanics computing component, and a fuel performance computing component; wherein, the neutron transport computing component and the thermal-hydraulic computing component achieve nuclear-thermal coupling simulation; the thermal-hydraulic computing component and the structural mechanics computing component achieve fluid-thermal-solid coupling simulation; the neutron transport computing component, the thermal-hydraulic computing component, and the fuel performance computing component achieve nuclear-thermal-fuel coupling simulation.

[0008] Furthermore, the coupler includes a grid mapping tool and a data conversion tool, and realizes the grid mapping and data conversion required for the coupling simulation between multi-physics core computing components through two coupling methods of operator splitting and Picard iteration.

[0009] Furthermore, the neutron transport computing component solves the neutron transport equation based on the direct three-dimensional characteristic line method, obtains the reactor temperature distribution and deformed geometric structure, generates / updates the cross-section data by combining the nuclide composition provided by the existing burnup model, and relies on the E-class heterogeneous supercomputer to allocate the computing tasks to the host side and the device side. The host side initializes the program grid, quadrature group, flat source region, and characteristic line, and divides the tasks according to the computing resources; the device side performs the transport solution of the characteristic line method for its respective task area and transmits the calculation results to the host side for convergence determination.

[0010] Furthermore, the thermal-hydraulic computing component includes a sub-channel module and a computational fluid dynamics module; the sub-channel module is used for the thermal-hydraulic simulation at the full-core level, supporting the large-scale parallel simulation of the full core of the pressurized water reactor quadrilateral assembly and the sodium-cooled fast reactor hexagonal assembly; the computational fluid dynamics module is used for the high-precision simulation of the physical field of the transient fluid in the reactor core, supporting two turbulence numerical simulation methods of direct numerical simulation and large eddy simulation, and solving the Navier-Stokes equation and the energy equation by using the high-order spectral element method.

[0011] Furthermore, the structural mechanics computing component is used for the static and fluid-induced vibration calculations of the reactor core components under the combined action of high temperature, irradiation, fluid action, and pressing force, and solves the control equation of structural mechanics by using the finite element method to obtain the bending deformation and vibration conditions of the core components.

[0012] Furthermore, the fuel performance computing component includes a microscopic scale simulation tool, a mesoscopic scale simulation tool, and a macroscopic scale simulation tool; wherein, the microscopic scale simulation tool includes a molecular dynamics module and an atomic dynamics Monte Carlo module, which are used to simulate the atomic motion and defect evolution at the microscopic level of nuclear fuel; the mesoscopic scale simulation tool includes a cluster dynamics module and a discrete dislocation dynamics module, which are used to analyze the defect aggregation and dislocation behavior at the mesoscopic scale of nuclear fuel; the macroscopic scale simulation tool obtains the thermodynamic and kinetic properties of nuclear fuel based on the force and thermal performance analysis tool.

[0013] Further, the support tool library includes: a basic mathematics library, a mesh generation tool, a V&V support tool, and a parallel I / O tool, where

[0014] The basic mathematics library includes a data structure module, a data primitive module, a basic operator module, a linear equation system solving module, and a unified call interface module, providing underlying mathematical calculation support for the core computing components;

[0015] The mesh generation tool parametrically generates tetrahedral meshes for the solid domain and fluid domain of the reactor core, and hexahedral meshes for the fluid domain;

[0016] The V&V support tool realizes the verification and validation functions of simulation results through a numerical simulation software test framework;

[0017] The parallel I / O tool provides performance diagnosis, performance models, and I / O strategy selection for large-scale parallel I / O for the neutron transport calculation component, the thermal-hydraulic calculation component, the structural mechanics calculation component, and the fuel performance calculation component.

[0018] Further, the numerical simulation software test framework includes a bottom-layer tool layer, an execution logic layer, and a test interface layer. Among them, the bottom-layer tool layer integrates a floating-point error analysis tool, a code debugging tool, a code management tool, and a code compilation tool; the execution logic layer provides the basic process for verifying and validating simulation results; the test interface layer encapsulates the interfaces for traditional testing, metamorphic testing, accuracy order analysis, regression testing, and error analysis.

[0019] Further, the simulation database includes a metadata management module and a simulation data module. The metadata management module uniformly describes the data of all calculation modules. The simulation data module includes a neutron transport data module, a thermal-hydraulic data module, a structural mechanics data module, and a nuclear fuel data module, which respectively store and manage the condition data, input parameters, and output results generated in the simulation of neutron transport, thermal-hydraulics, structural mechanics, and nuclear fuel performance of the nuclear reactor, and archive, version control, and trace the data.

[0020] Compared with the prior art, the present invention has the following advantages:

[0021] The present invention can not only support the single simulation and analysis of core physical processes such as reactor neutron transport, thermal hydraulics, structural mechanics, fuels and materials, but also support the refined coupled simulation of various physical processes. At the same time, the common parts of each core computing component are uniformly designed and implemented in the form of support tools, which improves the usability and functional scalability of the system, and provides the verification and validation functions for numerical reactor simulation, ensuring the accuracy and reliability of the simulation results. The proposed advanced numerical nuclear reactor system provides an economical and efficient test platform for the design optimization of existing pressurized water reactors and self-developed advanced third and fourth generation reactors, the simulation optimization of different operating conditions, the demonstration and prediction of severe accident sequences, and the research and development of fuels and materials. Description of the Drawings

[0022] The drawings generally illustrate various embodiments by way of example rather than limitation, and are used together with the description and the claims to explain the embodiments of the invention. Where appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the device or method.

[0023] Figure 1 It is the architecture diagram of the supercomputer numerical nuclear reactor system;

[0024] Figure 2 It is the schematic diagram of multi-physics coupling in the supercomputer numerical nuclear reactor system; among them, the black solid arrows indicate the parameter transfer directions between each computing component, and the black dotted lines indicate the coupler scheduling relationships;

[0025] Figure 3 It is the schematic diagram of the calculation process of the neutron transport computing component;

[0026] Figure 4 It is the schematic diagram of the calculation process of the sub-channel module of the thermal hydraulics computing component;

[0027] Figure 5 It is the schematic diagram of the calculation process of the CFD module of the thermal hydraulics computing component;

[0028] Figure 6 It is the schematic diagram of the calculation process of the fuel performance computing component;

[0029] Figure 7 It is the schematic diagram of the basic math library component module;

[0030] Figure 8 It is the schematic diagram of the calculation process of the mesh generation tool;

[0031] Figure 9 It is the schematic diagram of the NuSTA test framework;

[0032] Figure 10It is a schematic diagram of the simulation database component module. Detailed implementation manners

[0033] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will describe this application in detail with reference to the drawings and in combination with the embodiments.

[0034] This embodiment provides a supercomputer numerical nuclear reactor system for high-fidelity simulation of nuclear reactors, including a multi-physics simulation environment composed of a user interface, multi-physics core computing components and a coupler, a support tool library and a simulation database, and a hardware support environment. Among them,

[0035] The user interface is the top layer, composed of components directly interacting with users, including workflow management and visualization, etc., and is used to receive basic parameters such as reactor design parameters, operation parameters, accident condition parameters required for nuclear reactor simulation, as well as specific simulation requirements, etc.

[0036] The multi-physics simulation environment is the middle layer, which is the core part of the numerical nuclear reactor, including multi-physics core computing components and a coupler, and is used to implement the calculations of core physical processes such as neutron transport, thermal-hydraulics, structural mechanics, and fuel performance in nuclear reactors, as well as the coupled calculations of these physical processes.

[0037] The support tool library and the database are the sub-bottom layer. The support tool library is used to provide the required basic mathematical libraries, pre-processing mesh generation tools, V&V (Verification and Validation) support tools, and large-scale parallel I / O (Input / Output) tools, etc. for the calculations of various physical processes in the multi-physics simulation environment. The simulation database is used to store a large amount of simulation data for the multi-physics process simulation of nuclear reactors.

[0038] The hardware support environment is the bottom layer, mainly providing the required hardware environment resources for the numerical nuclear reactor system.

[0039] The coupler in the multi-physics simulation environment provides two coupling methods based on operator splitting and Picard iteration, and provides a mesh mapping tool and a data conversion tool to realize the mesh mapping and data conversion required for coupled simulation between multi-physics core computing components. Multi-physics coupling includes: nuclear-thermal coupling between neutron transport and thermal-hydraulic sub-channels, fluid-thermal-solid coupling between thermal-hydraulic CFD and structural mechanics, nuclear-thermal-fuel coupling between neutron transport and thermal-hydraulic CFD and fuel, etc.

[0040] The multi-physics core computing components in the multi-physics simulation environment include a neutron transport computing component, a thermal-hydraulic computing component, a structural mechanics computing component, and a fuel performance computing component. Among them,

[0041] The neutron transport calculation component is based on the direct three-dimensional method of characteristics (MOC) to solve the neutron transport equation, obtain the reactor temperature distribution and deformed geometric structure, generate / update cross-section data (absorption cross-section, scattering cross-section, fission cross-section, etc.) by combining the nuclide composition provided by the existing burnup model, and allocate the calculation tasks to the host side (i.e., CPU) and the device side (i.e., accelerator). The host side initializes the program grid, quadrature group, flat source region, and characteristics line, and divides the tasks according to the computing resources; the device side performs the transport solution of the method of characteristics for its respective task regions and transmits the calculation results to the host side for convergence determination. After the program execution is completed, data such as the neutron flux distribution and power distribution in each grid are stored in the database or transmitted to other calculation components through the coupler.

[0042] The thermal-hydraulic calculation component: includes two modules, namely the sub-channel and computational fluid dynamics (CFD). Among them, the sub-channel module is used for pin-by-pin thermal-hydraulic simulation at the full-core level, supporting large-scale parallel simulation of the full core of pressurized water reactor quadrilateral assemblies and sodium-cooled fast reactor hexagonal assemblies. The coolant flow channel surrounded by four rods (pressurized water reactor) or three rods (fast reactor) in the assembly is regarded as a sub-channel, and several sub-channel control volumes are divided axially for each sub-channel. The momentum conservation equation, mass conservation equation, and energy conservation equation are solved for each sub-channel control volume to obtain the flow field, pressure field, temperature field, void fraction distribution, and critical heat flux density distribution of the full core for full-core thermal-hydraulic characteristic analysis. This module first conducts sub-channel modeling of the full core, performs pin-by-pin sub-channel division for the core flow channels, and then divides the sub-channels into several solution domains according to the parallel task division algorithm, and allocates each solution domain to the corresponding processor for iterative solution of the control equations. After the program execution is completed, data such as the overall core pressure drop, temperature rise, and the flow rate and temperature of each sub-channel are stored in the database or transmitted to other calculation components through the coupler.

[0043] The CFD module is used to perform high-precision simulations of the physical fields of transient fluids in the reactor core. It supports two turbulence numerical simulation methods, namely Direct Numerical Simulation (DNS) and Large Eddy Simulation (LES). The Navier-Stokes equations and energy equations are solved using the high-order spectral element method to obtain the velocity, pressure, and temperature distributions of the physical fields. The use of high-order methods makes the solution results more accurate. This component takes the grid file of the reactor fluid domain obtained by the grid generation tool as input, distributes the computing tasks to the host and device sides. The host side initializes the communication library and performs fluid domain grid generation and processor mapping according to the computing resources. The device side constructs the right-hand side terms of the equation system and performs iterative solutions for their respective tasks, and then transmits the results back to the host side for output. After the program execution is completed, data such as the velocity, pressure, and temperature distributions within the grid are transmitted to other computing components.

[0044] The structural mechanics calculation component is used to perform static and fluid-induced vibration calculations of the reactor core components under the combined action of factors such as high temperature, irradiation, fluid action, and compaction force, to obtain the bending deformation and vibration conditions of the core components. This component uses the Finite Element Method (FEM) to solve the control equations of structural mechanics (including equilibrium equations, geometric equations, and physical equations). This component first calls the grid generation tool to obtain the solid domain grid of the core fuel rods and their assemblies, and then divides the spatial region according to the given hardware computing resources, and distributes the computing tasks to different computing nodes. Each computing node performs the finite element calculation process for its respective region, including: parallel derivation of the element stiffness matrix based on the grid of the current region, assembly of the element stiffness matrix into the total stiffness matrix of the current region, solution of the sparse stiffness matrix vector multiplication, and solution of the divided sub-equation systems. Among them, the operation of solving the sparse stiffness matrix vector multiplication is distributed to the device side, and the basic mathematical library in the underlying support tool library is called to complete the calculation. Other calculation processes are completed on the host side, and data communication between the device side and the host side, and between nodes is performed as needed.

[0045] The fuel performance calculation component is used to perform performance analysis of the irradiation damage of reactor nuclear fuel, including microscopic, mesoscopic, and macroscopic scale simulation tools. Among them, the microscopic scale includes Molecular Dynamics (MD) and Atomistic Kinetic Monte Carlo (AKMC) modules. The AKMC includes an on-lattice calculation model applicable to simple systems and an off-lattice calculation model applicable to complex systems; the mesoscopic scale includes Cluster Dynamics (CD) and Discrete Dislocation Dynamics (DDD) modules. Among them, the CD includes a deterministic CD calculation model applicable to simple systems and a stochastic CD calculation model applicable to complex systems; on the macroscopic scale, the thermodynamic and kinetic properties of nuclear fuel are obtained based on a force and heat performance analysis tool. The MD module first reads the PKA energy spectrum to simulate the generation process of initial defects, obtaining the distribution of damaged atoms. This distribution is used as the input of the AKMC module. The AKMC selects an appropriate calculation model according to the complexity of the simulation system to realize the nucleation and growth simulation of initial defects. Among them, the distribution of damaged atoms obtained by MD simulation can use the intelligent analysis tool SACD to realize the statistical and clustering analysis of initial defects, obtaining the size and spatial distribution of defects such as vacancies, self-interstitials, and their clusters. The results are stored in the Cascade database and used as the input for AKMC, CD, DDD, etc. Then, the CD module selects an appropriate calculation model according to the complexity of the simulation system to realize the long-term evolution simulation of defects, obtaining information such as the size, number density, and spatial distribution of irradiation defects. Then, the DDD module directly simulates the interaction between dislocations and defects according to the defect distribution, and obtains mechanical parameters (temperature, stress, strain, etc.) after long-term evolution and inputs them into the macroscopic scale analysis tool to realize the mechanical performance analysis and prediction of nuclear fuel. Each scale module realizes parallel solution of the dynamic equation at the host end through domain decomposition and communication, and parallel acceleration is realized at the heterogeneous device end (acceleration card) for calculation hotspots such as potential functions and defect interaction forces. Among them, the defects are nuclear fuel fission gases and solid fission products, etc.

[0046] The support tool library includes a basic mathematics library, a mesh generation tool, a V&V support tool, and a parallel I / O tool. Among them,

[0047] The basic mathematics library is used to provide underlying mathematical calculation support for the core calculation components in the multi-physics coupling system, including: data structure module, data primitive module, basic operator module, linear equation system solving module, and unified call interface module. Among them, the data structure module provides a unified data storage format, the data primitive module provides a heterogeneous communication interface for basic operators on domestic supercomputer acceleration cards, the basic operator module provides sparse operator and dense operator solving functions, the linear equation system solving module provides preconditioners and iterative solving algorithms for large-scale linear equation systems, and the unified call interface module provides interface calls for sparse operators, dense operators, large-scale linear equation system solving, and large-scale equation solving preconditioning during large-scale parallel simulations for each module of the multi-physics coupling system, realizing the function call of the multi-physics coupling system to the basic mathematics library.

[0048] The mesh generation tool is used to provide the required meshes for the high-fidelity simulation calculation of the reactor. It can parametrically generate tetrahedral meshes for the solid domain and fluid domain of the reactor core and hexahedral meshes for the fluid domain. The mesh generation tool receives the reactor parameters provided by the user, constructs the geometric model of the corresponding component unit segment, and then generates tetrahedral meshes or hexahedral meshes for the component unit segment according to the user's requirements for the mesh type. Among them, for the generation of tetrahedral meshes of the component unit segment, a commercial software is first used to generate the basic mesh, and then the basic mesh is refined; for the generation of hexahedral meshes of the component unit segment, the differential idea is mainly adopted. First, the surface meshes of several z-axis cross-sections of the component unit are generated, and then the surface meshes are rotated and translated to generate all the surface meshes of the component unit segment. Then, the corresponding surface meshes on each cross-section are connected to form the hexahedral mesh of the component unit segment. Finally, the generated meshes of the component unit segment are copied and translated according to the input core parameters, and the tetrahedral meshes of the solid domain and fluid domain of the core and the hexahedral meshes of the fluid domain are generated in parallel.

[0049] The V&V support tool provides a numerical simulation software testing framework NuSTA, which provides the verification and validation functions for the simulation results of the multi-physics coupling system. This framework includes an underlying tool layer, an execution logic layer, and a test interface layer. The underlying tool layer integrates common floating-point error analysis tools, code debugging tools, code management tools, and code compilation tools to provide basic software testing functions; the execution logic layer provides the basic processes for verifying and validating simulation results, and each process realizes the corresponding functions by calling the underlying testing tools; the test interface layer encapsulates the interfaces of functions such as traditional testing, metamorphic testing, accuracy order analysis, regression testing, and error analysis, providing various software testing interface calls for verification and validation.

[0050] The parallel I / O tool is used to provide performance diagnosis, performance models, and I / O strategy selection for large-scale parallel I / O of core computing components (such as structural mechanics, thermal hydraulics, neutron transport, fuel, and couplings, etc.). It mainly includes: ① Measuring the I / O performance of the input data of a single component under different computing resources and different I / O strategies; ② Based on the isomorphic properties of core components, establishing an I / O performance model under multiple components-computing resources-multiple I / O strategies; ③ Given resources and computing cases, selecting the optimal I / O based on the I / O performance model.

[0051] The simulation database is used to store, manage, and share the input, output data, and their relationships of each computing component in the multi-physics simulation of the reactor. The simulation database mainly includes a metadata management module and a simulation data module. Among them, the metadata management module uniformly describes the data of all computing modules, including meta-information such as data format, unit, range, source, etc., and provides a case identifier (CaseID) for associating the input and output data of different modules, supporting data annotation and classification for quick retrieval and version management; the simulation data module includes a neutron transport data module, a thermal hydraulics data module, a structural mechanics data module, and a nuclear fuel data module. Each contains three sub-modules: a case module, an input module, and an output module. Among them, the neutron transport data module records the case information, input information, and calculation result information including power distribution for supporting the neutron transport simulation of the reactor physics field; the thermal hydraulics data module records the case information, input information, and calculation result information such as temperature field and flow velocity field for the accurate simulation of reactor coolant flow and heat transfer; the structural mechanics data module records the case information, input information, and calculation result information including stress distribution and structural deformation for the mechanical property simulation of the reactor structure; the nuclear fuel data module records the case information, input information, and calculation result information including irradiation damage evolution for simulating the defect evolution and mechanical properties of nuclear fuel under irradiation conditions.

[0052] Embodiment 1

[0053] Figure 1 The figure shows a schematic diagram of the supercomputer advanced numerical nuclear reactor system architecture of this embodiment.

[0054] As Figure 1As shown in the figure, the numerical nuclear reactor system includes four levels. The top-level user interface consists of components that directly interact with users, including workflow management and visualization, etc., and is used to receive basic parameters such as reactor design parameters, operating parameters, accident condition parameters required for nuclear reactor simulation, as well as specific simulation requirements, etc.; the next-level multi-physics simulation environment is the core layer of the system, and is used for independent simulation and coupled simulation of core physical processes such as neutron transport, thermal-hydraulics, structural mechanics, and fuel performance, including core computing components such as neutron transport calculation components, thermal-hydraulic calculation components, structural mechanics calculation components, and fuel performance, as well as a multi-physics coupler. The multi-physics coupler realizes the coupling of each physical process based on two coupling methods: operator splitting and Picard iteration, realizes the scheduling of each computing component during the coupled simulation process through a task scheduler, and provides grid mapping tools and data conversion tools required for the coupled simulation; the sub-bottom support tool library and database provide common support tools for the multi-physics simulation environment, including basic mathematics libraries, pre-processing grid meshing tools, V&V support tools, and parallel I / O tools. The database is used to store and manage the massive simulation data generated by each core computing component in the multi-physics simulation environment; the hardware support environment at the lowest level is mainly provided by national E-class supercomputers such as Shenwei Ocean Light, the new generation of Tianhe Supercomputer, and the new generation of "CPU + DCU" supercomputers, and is used to provide the required hardware environment resources for the numerical nuclear reactor system.

[0055] Figure 2 This is the multi-physics coupling schematic diagram of this embodiment. After the coupler obtains reactor parameters (such as reactor shape, nuclear power, coolant inlet and outlet temperatures, etc.) and coupled simulation requirements through the user interface, it selects a coupling method based on operator splitting or Picard iteration, and realizes the coupled simulation of different physical processes such as neutron transport, thermal-hydraulics, structural mechanics, and fuel performance through coupled scheduling. Specifically, it includes: realizing nuclear-thermal coupled simulation through the neutron transport component and the thermal-hydraulic sub-channel component, realizing fluid-thermal-solid coupled simulation through the thermal-hydraulic CFD component and the structural mechanics component, realizing nuclear-thermal-fuel coupled simulation through the neutron transport component, the thermal-hydraulic CFD component, and the fuel performance component, etc., and calling the grid mapping tool and the data conversion tool to realize the grid mapping and data format conversion between different computing components.

[0056] Next, taking the new generation of E-class "CPU + DCU" supercomputer as an example, the specific implementation methods of the multi-physics core computing components, support tool library, and database in this embodiment are described, where the CPU side is the host side and the DCU is the device side.

[0057] (1) Neutron transport calculation component

[0058] The calculation process of the neutron transport calculation component described in this embodiment is as Figure 3As shown in the figure, according to the characteristics of the "CPU + DCU" heterogeneous architecture, data reading, output, program initialization, and calculation task division are allocated to the CPU side, and the calculation and solution of the neutron transport equation are allocated to each DCU side for execution. After obtaining the temperature distribution and deformed geometric structure, cross-section data and reactor geometric structure are respectively constructed on the CPU side; in the program initialization stage, the subdivision of rings and sectors in the geometric structure, the initialization of quadrature set weights, the initialization of materials in the flat source region, and the initialization of characteristic lines are completed; for the characteristics of the "CPU + DCU" heterogeneous architecture, the calculation tasks are divided into each DCU side, and through asynchronous data transmission, the data required by each node is transmitted to the DCU side to complete the data transmission from the CPU side to the DCU side; the DCU side reads data such as trajectories, line segments, and fluxes in the video memory according to its own calculation task division, and executes the characteristic line method transport solution module, in which kernel functions such as source terms, angular fluxes, and scalar fluxes are executed in sequence; to accelerate the transport solution, a theoretical acceleration module - coarse mesh finite difference acceleration is used to convert the neutron transport equation into a neutron diffusion equation required for coarse mesh finite difference calculation and perform iterative solution; the flat source region flux and k eff and other key data updated in the current iteration round are passed back to the DCU video memory and then transmitted back to the CPU side; the CPU side is responsible for the convergence determination of the calculation results returned by the DCU side. First, it receives the neutron transport results calculated by the DCU side, including neutron flux distribution, k eff and related physical quantities. During the convergence determination process, the CPU side compares the current calculation results with the previous iteration results and judges whether the convergence conditions are met according to the preset convergence criteria; after the CPU side determines that the program calculation results converge, it transmits data such as power distribution and neutron fluence rate required by other calculation components; key data during the program execution, such as the scalar flux of each flat source region, each characteristic line, the starting coordinates of trajectory segments, and angular fluxes in each iteration round, are stored in the simulation database.

[0059] (2) Thermal-hydraulic calculation component

[0060] The calculation process of the sub-channel calculation module described in this embodiment is as Figure 4 shown. According to the characteristics of the supercomputer heterogeneous architecture (taking the "CPU + DCU" heterogeneous architecture as an example), the establishment of the pressure coefficient matrix and the iterative solution of the pressure coefficient matrix in the solution process are allocated to each DCU side for execution, and the rest are allocated to the CPU side for execution.

[0061] 1) First, perform full-core sub-channel modeling according to the user's simulation requirements. The sub-channel modeling of PWR quadrilateral assemblies and the sub-channel modeling of fast reactor hexagonal assemblies can be selected according to the user input.

[0062] 2) Perform parallel task division on the results of the full-core sub-channel modeling, and divide the full-core sub-channels into several solution domains.

[0063] 3) Parallelly solve the parallel tasks generated by the parallel task division part, and allocate one CPU core for each solution domain for solving. When solving, first solve the solid heat conduction, traverse all the rods to calculate the temperature distribution inside the rods, traverse all the sub-channels in the solution domain and calculate the heat transfer coefficient. Subsequently, establish and solve the momentum conservation equation to obtain the velocity, momentum, and mass flow rate of each sub-channel control volume. Finally, establish the mass and energy conservation equations and call the DCU to solve them.

[0064] 4) DCU solution part: First, perform data transmission to transfer parameters such as pressure, void fraction, density, and mass flow rate required for the solution to the DCU video memory. Then, construct the pressure coefficient matrix according to the process of establishing the mass and energy conservation equations on the CPU side. Use the iterative method to solve the pressure coefficient matrix. Finally, transfer the solution result back to the host side and perform the calculation of the next time step.

[0065] The calculation process of the CFD calculation module described in this embodiment is as Figure 5 shown. According to the supercomputer heterogeneous architecture, tasks such as input of operating parameters, output, program initialization, and calculation task division in the CFD calculation module are allocated to the CPU side, and the solution of the velocity and pressure equations is allocated to each DCU side for execution. On the CPU side, during the input stage of operating parameters, parse the fluid domain grid file and perform regional decomposition, and set the control parameters required for the calculation; during the solver initialization stage, it is necessary to parse the geometric structure of the fluid domain grid, construct the geometric matrix, construct the constants related to the spectral element method, and initialize the communication library at the same time; according to the regional decomposition result, the CPU side divides the calculation tasks into each DCU side, and through asynchronous data transmission, transfers the data required by each node to the DCU side to complete the data transmission from the CPU side to the DCU side; the DCU side accelerates the core part of the calculation, reads relevant information from the DCU video memory, and constructs the right-hand side terms of the equations, including the right-hand side term of the pressure equation, the right-hand side term of the velocity equation, and the explicit term F; during the iterative solution of the linear equation, calculate the residual and define the Krylov subspace, and iteratively find the best approximation in the affine space as the calculation result and transfer it back to the CPU side. The CPU side can transfer the data of the obtained pressure, velocity, and temperature distribution to other calculation components.

[0066] (3) Structural mechanics calculation component

[0067] The structural mechanics calculation component in this embodiment first calls a mesh generation tool to obtain the solid domain mesh of the core fuel rods and their assemblies, and then divides the spatial region (i.e., the solid domain mesh) according to the given hardware computing resources, and distributes the calculation tasks to the corresponding computing nodes. Each computing node executes the finite element calculation process for its respective region (i.e., uses the finite element method to solve structural mechanics control equations such as the equilibrium equation, geometric equation, and physical equation): First, on the CPU side, the parallel element stiffness matrix is derived based on the solid domain mesh of the current region to obtain the element stiffness matrices of each sub-region; then, the element stiffness matrices of each sub-region are assembled into the global stiffness matrix of the current region; next, the global stiffness matrix information on the CPU side is sent to the DCU side, and the sparse stiffness matrix vector multiplication is solved on the DCU side. This operation is completed by calling the basic mathematics library in the underlying support tool library, and the calculation result is sent back to the CPU side; finally, the divided support equation set is solved on the CPU side to obtain the mechanical behaviors such as the bending deformation and vibration of the fuel assembly.

[0068] (4) Fuel performance calculation component

[0069] The calculation process of the fuel performance calculation component described in this embodiment is as Figure 6 shown. This component uses parallel simulation tools (MD, AKMC, CD, DDD, etc.) at different space-time scales to achieve multi-scale coupling simulation at the microscopic and mesoscopic scales. Among them,

[0070] 1) Parallel MD realizes the simulation of initial defect generation

[0071] In this embodiment, the spatial simulation region is first divided so that each CPU process is independently responsible for one region. After the CPU reads the PKA energy spectrum, it transmits the energy spectrum information and atomic attribute information to the DCU side, and the potential energy between atoms is calculated in parallel on the DCU side; then the potential energy is transmitted from the DCU side to the CPU side and the atomic force, velocity, displacement, etc. are continued to be calculated; to alleviate the huge data transmission and communication bottleneck, block data storage is performed based on local shared memory, and the potential table communication and potential function calculation are overlapped to cover the transmission overhead based on the double-buffer method of stream technology; finally, multiple time steps are iteratively looped to obtain the initial damaged atom distribution. And based on the intelligent analysis and statistical tool SACD, the damage information is stored in the cascade defect library.

[0072] 2) Parallel AKMC realizes the simulation of defect nucleation and growth

[0073] After receiving the initial damaged atom distribution information, AKMC simulates the nucleation and growth behaviors of simple and complex defects based on the on-lattice and off-lattice calculation models respectively. The on-lattice and off-lattice divide the simulation area into equal volumes to achieve CPU parallel computing. Each CPU calculates the vacancy transition probability, event selection, and execution of event transitions in sequence within a single time step. For the off-lattice module of complex defects, reaction events are first obtained through the saddle point search algorithm at the CPU end and an event table is constructed. The event table is transmitted to the DCU end so that each thread evenly distributes multiple reaction event search tasks; the search results are sent back to the CPU end to continue calculating the atomic forces and energies. Among them, AKMC and MD use the DCU end to accelerate the potential function calculation to calculate the atomic forces, transition energies, and transition probabilities in sequence, and then execute the reaction event with the highest probability.

[0074] 3) Parallel CD realizes the long-term evolution of defects

[0075] This implementation receives the defect distribution obtained by MD and / or AKMC, and selects the deterministic CD or stochastic CD calculation model according to the complexity of the simulation system to perform the long-term evolution process of defects in the corresponding system. First, it realizes the spatial region decomposition with MD and AKMC, and each CPU independently executes the calculation tasks of the current region. For the deterministic CD calculation model for simple systems, the calculation task of the current region is to solve the CD equation, which is solved by the implicit iterative method based on the Jacobian matrix. Further, the construction task of the Jacobian matrix is assigned to the DCU end for execution, and finally the overall solution of the CD equation is carried out at the CPU end; for the stochastic CD calculation model for complex systems, the calculation task of the current region is the selection and update of defect reactions, which is only carried out at the CPU end. The defect size, number density, and spatial distribution are obtained by iterating the time step.

[0076] 4) Parallel DDD realizes the simulation of the interaction between dislocations and defects

[0077] DDD receives the long-term defect evolution information, divides the spatial region with MD, AKMC, and CD, and enables each CPU to independently solve the dislocation dynamics equation. Among them, the fast multipole method based on heterogeneous parallelism is used to solve the far-field dislocation interaction, and the near-field interaction force is accelerated by fitting based on a deep neural network. The implicit time integration strategy is used to select the time step and update the positions of the dislocation nodes. The regional boundaries are adaptively adjusted according to the execution time of each CPU to achieve polymorphic load balancing. Information such as the system temperature, stress, and total strain is obtained through iteration.

[0078] Finally, information such as stress and strain is transmitted to the macroscopic performance analysis tool, and the irradiation damage mechanical properties of nuclear fuel are predicted accordingly. Among them, the defects are nuclear fuel fission gases and solid fission products, etc.

[0079] (5) Support Tool Library

[0080] 1) Basic Mathematics Library

[0081] Figure 7 For the overall design of the basic mathematics library components described in this embodiment, the following are the designs of each module and their core functions:

[0082] i. Data Structure Module: Use the data of each iteration step in the iterative solution of the multi-physical coupling system to fill the basic data structure, complete the data transfer from the CPU side to the DCU side, and complete the data preparation before basic calculations;

[0083] ii. Communication Primitive Module: Based on the characteristics of the DCU architecture, design and implement multi-communication primitives (prefix sum, reduction, sorting) at different levels (thread block level, warp level, thread level) to complete the communication operations between DCU multi-threads under various precision types;

[0084] iii. Basic Operator Module: It includes two parts: sparse operator and dense operator. Integrate the existing algorithms of each operator, design an algorithm adaptive strategy selection, select the number of non-zero elements in the matrix, the matrix sparsity, and the average number of non-zero elements per row of the matrix as matrix features, and match corresponding solution algorithms for different feature matrices. Deeply optimize different algorithms according to the architecture characteristics, and achieve an average distribution of the computational workload of each operator on the DCU hardware through the cooperation of block-level load balancing and thread-level load balancing; Use LDS as the intermediate result storage medium to improve the memory access efficiency of the algorithm; The prefix sum, reduction, and sorting operations in the algorithm are implemented by calling the interfaces at each level in the communication primitive module;

[0085] iv. Linear Equation Solving Module: Integrate the preconditioners and iterative algorithms required in the multi-physical coupling system. The basic operators and DCU communication in the algorithm flow are implemented by calling the interfaces of the basic operator module and the communication primitive module;

[0086] v. Unified Call Interface Module: It includes the interfaces of each operator in the basic operator module and the interfaces of each preconditioner and iterative algorithm in the linear equation solving module, for the multi-physical coupling system to call the solution function of the basic mathematics library module.

[0087] 2) Mesh Generation Tool

[0088] The calculation process of the mesh generation tool described in this embodiment is as Figure 8 shown. Here, the China Experimental Fast Reactor CEFR is taken as an example to illustrate the specific implementation method.

[0089] i. The reactor parameter reading module obtains the geometric information (such as fuel rod radius, wire winding radius, component position, etc.) and mesh information (such as mesh scale, etc.) of the CEFR core from the user;

[0090] ii. According to the obtained geometric information, by defining the dimensions, shapes, and positional relationships of the geometric model, construct the geometric model of a component unit section with a pitch (or a given axial length) in the axial direction as the geometric model of the component unit section;

[0091] iii. According to the generated geometric model of the component unit section, divide the tetrahedral mesh or hexahedral mesh of the component unit section according to the user's grid type requirements, where,

[0092] a) Generation of the tetrahedral mesh of the component unit section. By using the isomorphic characteristics of the reactor, a highly refined and high-quality grid model of the core flow field can be constructed. The construction process is divided into three stages:

[0093] ① Rapidly construct the basic grid model of the component unit section with the help of commercial grid generation tools (such as GMSH, ICEM, etc.);

[0094] ② Carry out iterative refinement and optimization calculations on the basic grid model to obtain a high-quality grid model;

[0095] ③ Stitch the obtained high-quality unit section grids to construct the basic grid unit covering the components and gaps.

[0096] b) Generation of the hexahedral mesh of the component unit section. Cut the component unit section into numerous "layers" along the z-axis. Starting from the bottom of the component, divide it into n segments according to a certain angle of rotation of the wire winding for each segment. Divide several blocks in each layer of the first segment and generate grid points within the blocks. With the help of the rotational symmetry of the hexagon, rotate the grids of each layer of the first segment by 60° in sequence to generate the quadrilateral grids of each layer of the basic component section. Connect the corresponding quadrilateral grids of each layer to construct the hexahedral mesh model of the basic component section.

[0097] iv. Due to the axial and radial isomorphism of the CEFR core, perform geometric transformations such as translation, rotation, and mirroring on the component unit section along the axial and radial directions, and generate the grid models of the solid domain and fluid domain of the entire core in parallel. The final grid model is sliced into several small grid files and stored on the disk.

[0098] 3) V&V Support Tools

[0099] The V&V support tools described in this embodiment include code verification, numerical verification, solution verification, etc. in the V&V verification process, and can complete the corresponding test functions by calling the relevant interfaces of the NuSTA test framework test interface layer shown below. The specific process is as follows: Figure 9 The relevant interfaces of the NuSTA test framework test interface layer to complete the corresponding test functions. The specific process is as follows:

[0100] i. Determine the software to be verified, the parameters to be verified, the types to be verified, select the software test type, and call the corresponding test interface in the NuSTA test framework;

[0101] ii. Automatically generate relevant test cases through the execution logic layer test case definition module according to the selected software to be verified, parameters, and types.

[0102] iii. Generate an automatic test script through the execution logic layer test script generation module according to the large-scale parallel simulation process of the software to be verified and the test cases prepared in the previous step.

[0103] iv. Run the execution logic layer test driver, assemble the test process through the verification type and test type, and call the tools in the underlying tool layer to complete the test.

[0104] v. Call the test result analysis tool (valgrind) in the underlying tool layer to analyze the test results, and determine whether the test results pass through the test language machine.

[0105] vi. Determine whether to execute the regression test engine according to the test type selected in step i. If so, run the execution logic layer regression test engine to carry out regression test analysis.

[0106] vii. Store and manage the test result data through the reactor simulation database.

[0107] 4) Parallel I / O tool

[0108] For the parallel I / O tool described in this embodiment, first measure the I / O performance under a single component: Given a small computing resource set R1 = {p 1 , p 2 , …, p n}, a set of computing components (referring to structural mechanics, thermal hydraulics, neutron transport, fuel performance, and couplers, etc.) C = {C 1 , C 2 ,..., C m}, and an I / O strategy S = {S 1 , S 2 , …, S k}. Based on the single-component example, for various I / O strategies under each computing component C i (where i = 1, 2, …, m), carry out I / O performance tests under small computing resources to obtain performance data records (such as time). That is, for the computing component C i , there is a performance data record set Y = {y 1 , y 2 , …, y s}, where y j = Test(R1 × S), (where i = 1, 2, …, s; × represents the Cartesian product of sets, and Test represents the I / O performance test record under a specific computing strategy and resources).

[0109] Then, establish the I / O performance model under multiple reactor components: For the computing component C i , based on the single-component performance record set Y, establish the I / O performance model Model s (N, P) for each computing strategy s under multiple components, where N is the number of components and P is the number of computing resources (such as the number of MPI processes). Specifically, (Here is a linear model, and Model s (N, P) can also be other regression models).

[0110] Finally, perform strategy selection: Given the number of components N and the number of computing resources P corresponding to the computing task to be calculated, based on the model Model s (N, P), give the optimal I / O strategy for the computing component C i . According to the given optimal strategy, modify the strategy configuration (such as configuration files, compilation options, etc.) of the computing component C i , and execute the reactor I / O computing task.

[0111] (6) Simulation database

[0112] Figure 10 FIG. is the overall architecture diagram of the simulation database described in this embodiment, including a metadata management module and a simulation data module. This database is designed based on the document storage of a non-relational database (NoSQL), and uses a JSON document structure similar to MongoDB for storage. It standardizes and stores the operating conditions, input, and output data of each component module to ensure data integrity and consistency, avoid redundancy and conflicts, and at the same time support the archiving, version control, and tracking of all data during the simulation process, facilitating later analysis, debugging, and optimization. The following are the data field designs and core functions of each module, where the data type is its essential or representative data type, and the actual storage type varies according to the data of different computing components.

[0113] 1) Metadata module

[0114] Each metadata record corresponds to an input / output data item, and its fields include:

[0115] Field Name Type Description DataID String Data unique identifier, used to identify data items CaseID String Operating condition unique identifier, associated with specific operating conditions ModuleName String Name of the module to which the data belongs (such as neutron transport, thermal-hydraulics) DataType Enumeration type Data type (input or output) SimulateMethod String Simulation method (such as sub-channel, CFD, MD, DD, KMC) PhysicalQuantity String Physical quantity of the data (such as power distribution, temperature field) Unit String <![CDATA[Data units (such as W / cm 3 , K)]]> DataFormat String Data format (such as JSON, HDF5, CSV) GridType String Grid type (such as structured grid, unstructured grid) Source String Data source (such as calculation module name, experimental data) Timestamp Timestamp Data generation or modification time Version String Data version number, used for tracing and comparison Description String Data description (such as calculation conditions or background description)

[0116] 2) Simulation data module

[0117] To achieve cross-module data consistency, the following is the design of the same fields in the four data modules:

[0118] i. Unified fields in the operating condition module

[0119] Field Name Data type Description CaseID String Operating condition unique identifier, used to associate input and output modules SimulationMode Enumeration type Simulation mode (such as steady state, transient state, accident condition) ReactorGeometry String / JSON Geometric description (such as CAD model path or grid file path) BoundaryConditions JSON Boundary condition definition (general description interface) Timestamp Timestamp Operating condition creation or modification time Description String Operating condition description (such as operating conditions or background description)

[0120] ii. Unified fields of the input module

[0121] Field Name Data type Description InputID String Input data unique identifier CaseID String Associated operating condition identifier MeshType String Grid type (such as uniform grid, non-uniform grid) MeshData JSON Grid information (node coordinates and element connection relationship) InitialConditions JSON Initial conditions (general definition, such as initial field distribution) Timestamp Timestamp Input data generation or modification time Description String Input data description (such as boundary conditions or initial parameter description)

[0122] iii. Unified fields of the output module

[0123]

[0124]

[0125] The following is the design of the differential fields among the four data modules:

[0126] i. Neutron transport data module

[0127]

[0128] ii. Thermal-hydraulic data module

[0129]

[0130] iii. Structural mechanics data module

[0131]

[0132] iv. Fuel performance data module

[0133]

[0134]

[0135] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. A numerical nuclear reactor system based on supercomputing, characterized in that: The system includes a user interface, a multi-physics simulation environment, a support tool library and a simulation database, and a hardware support environment; the user interface is used to receive parameters and simulation requirements required for nuclear reactor simulation; the multi-physics simulation environment contains multi-physics core computing components and couplers, which are used to realize nuclear reactor core physical process calculations and coupling calculations; the support tool library provides support for each physical process calculation; the simulation database is used to store simulation data; and the hardware support environment is used to provide hardware resources.

2. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The multi-physics core computing component includes: a neutron transport computing component, a thermal-hydraulic computing component, a structural mechanics computing component, and a fuel performance computing component; wherein the neutron transport computing component and the thermal-hydraulic computing component realize nuclear-thermal coupling simulation; the thermal-hydraulic computing component and the structural mechanics computing component realize fluid-heat-solid coupling simulation; the neutron transport computing component, the thermal-hydraulic computing component, and the fuel performance computing component realize nuclear-heat-fuel coupling simulation.

3. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The coupler includes a grid mapping tool and a data conversion tool, and realizes the grid mapping and data conversion required for coupling simulation between multi-physics core computing components through two coupling methods of operator splitting and Picard iteration.

4. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The neutron transport calculation component solves the neutron transport equation based on the direct three-dimensional characteristic line method, obtains the reactor temperature distribution and deformation geometry, generates / updates cross-sectional data in combination with the nuclide composition provided by the existing burnup model, and distributes the calculation tasks to the host side and the device side based on the E-class heterogeneous supercomputer. The host side performs program grid, quadrature group, flat source area and characteristic line initialization, and divides the tasks according to the calculation resources; The device side performs characteristic line method transport solution for each task area and transmits the calculation results to the host side for convergence judgment.

5. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The thermal-hydraulic calculation components include a sub-channel module and a computational fluid dynamics module; the sub-channel module is used for full-core-level thermal-hydraulic simulation, and supports large-scale parallel simulation of the full core of pressurized water reactor quadrilateral components and sodium-cooled fast reactor hexagonal components; the computational fluid dynamics module is used for high-precision simulation of the physical field of transient fluid in the reactor core, supports two turbulence numerical simulation methods, direct numerical simulation and large eddy simulation, and uses high-order spectral element method to solve the Navier-Stokes equations and energy equations.

6. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The structural mechanics calculation component is used for calculating the statics and flow-induced vibration of the reactor core assembly under the combined effects of high temperature, irradiation, fluid action and compression force, and adopts the finite element method to solve the control equations of structural mechanics to obtain the bending deformation and vibration of the core assembly.

7. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The fuel performance calculation component includes a microscale simulation tool, a mesoscale simulation tool and a macroscale simulation tool; wherein the microscale simulation tool includes a molecular dynamics module and an atomic dynamics Monte Carlo module, which are used to simulate the atomic motion and defect evolution at the microscopic level of nuclear fuel; the mesoscale simulation tool includes a cluster dynamics module and a discrete dislocation dynamics module, which are used to analyze the defect aggregation and dislocation behavior of nuclear fuel at the mesoscopic scale; the macroscale simulation tool obtains the thermodynamic and kinetic properties of nuclear fuel based on the thermomechanical performance analysis tool.

8. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The support tool library includes: basic mathematics library, meshing tool, V&V support tool and parallel I / O tool, among which, The basic mathematical library includes a data structure module, a data primitive module, a basic operator module, a linear equation solving module, and a unified call interface module, which provide underlying mathematical computing support for the core computing components; The meshing tool generates parameterized tetrahedral meshes of the solid domain and fluid domain of the reactor core and hexahedral meshes of the fluid domain; The V&V support tool implements the verification and confirmation function of the simulation results through the numerical simulation software testing framework; The parallel I / O tool provides large-scale parallel I / O performance diagnosis, performance model and I / O strategy selection for the neutron transport calculation component, thermal hydraulic calculation component, structural mechanics calculation component and fuel performance calculation component.

9. The supercomputing-based numerical nuclear reactor system according to claim 8, characterized in that: The numerical simulation software testing framework includes an underlying tool layer, an execution logic layer, and a test interface layer, wherein the underlying tool layer integrates floating-point error analysis tools, code debugging tools, code management tools, and code compilation tools; the execution logic layer provides a basic process for simulation result verification and confirmation; the test interface layer performs interface encapsulation for traditional testing, degradation testing, precision order analysis, regression testing, and error analysis.

10. The supercomputing-based numerical nuclear reactor system according to claim 1, characterized in that: The simulation database includes a metadata management module and a simulation data module. The metadata management module uniformly describes the data of all calculation modules. The simulation data module includes a neutron transport data module, a thermal hydraulic data module, a structural mechanics data module and a nuclear fuel data module, which respectively store and manage the operating data, input parameters and output results generated by the nuclear reactor in the neutron transport, thermal hydraulics, structural mechanics and nuclear fuel process simulation, and archive, version control and track the data.

Images