Development And Parallel Optimization Of KMC Numerical Simulation Program For Radiation Damage Of Metal Materials

The positive development of nuclear power has brought an effective solution to the problem of energy consumption.The service life of the nuclear materials(steel pressure vessels-metal materials)that ensure the normal operation of nuclear reactors directly affects the service capacity of nuclear power plants.Main cause of the metal material performance degradation is its internal particle under irradiation condition and high energy incident particles collide(the microscopic changes,called irradiation damage),to predict material internal structure change process,this article will kinetic monte carlo method and the combination of high performance computing technology in China,the irradiation damage of the metal material for KMC relevant numerical simulation program for the development and the parallel optimization,specific work is as follows:(1)Dynamics theory,the basis of monte carlo method is studied based on the system evolution rules of this method are classified,and made a brief introduction to the related technical support platform(parallel computing platform architecture and visual simulation software),combining with the characteristics of AKMC algorithm: the transition probability calculation,random selection and calculation process on parallel numerical simulation program MISA-KMC is implemented and then a detailed introduction of main technical involved(including parallel domain decomposition,second row molecular domain and the method of synchronous communication between processes,etc.).(2)To relevant optimization and performance analysis of software,introduced a kind of calculation model of atomic potential energy(Pair potential),reduce the computational overhead of energy in the simulation process,improve the efficiency of calculation,and on this basis,extending the application of simulated atomic species realize the evolution of the multielement alloy,finally using visual analysis to the simulation result is studied.By using the character of atomic coordinates in the program,the isolated atom recognition function is realized,which guarantees the correctness of the simulation results.The local timing method is used to analyze the hot spot function in the program to provide a solution to adapt to the architectural characteristics of the supercomputer.The preset parameters in the program input file are adjusted to optimize the operation state while ensuring the correct simulation results.The program was transplanted to "sunway · taihulight" for performance analysis to verify the feasibility of the optimization method and the scalability of the parallel program.A fixed synchronization interval algorithm is implemented to solve the problem of data loss in some cases.(3)Based on the increasing importance of fusion materials,F-S potential function selection mode and W-Re binary alloy(fusion material structure)system are added in the program.The feasibility of the method is verified by calculating part of the energy value,and the MISA-KMC numerical simulation program is extended in the application field.